Although, this is an old and dusted topic the points I intend to discuss here are based on my 7 years of experience with mobile app development.<img src="https://media.coffeebeans.io/blogs/Hybrid-Vs-Native-App-1-2048x1152.jpg" alt="cover">Hybrid vs Native Battle
Yes, no doubt that a Native app can apparently give a better user experience. And no doubt that if you have budget constraints going native is a challenge. But there are a few points that we need to consider before making the decision. Especially, when there are so many beautiful hybrid app development platforms now.
<ol>
<li>What is the purpose of the app? If the app is a utility and adds to the existing business you are doing, it is worth considering a hybrid app. If the app itself is the main business, think if you really need a native app? I suggest native only if you have to use a lot of native features like GPS, the user experience needs to be kickass, and if you have the money. Most of the banking apps have gone for hybrid, to give you an example.</li>
<li>**Cost of ownership. **There is no doubt that the cost of development for native is almost double considering multiple platforms. But think about the total lifecycle cost. Each update to the OS, especially with iOS, needs reworking of your code. There might be deprecations, etc. that you will have to fix even if there are no functionality changes. Overall, the cost of maintenance is high.</li>
<li>**Time to Market. **Hybrid or Mobile app solution is a great way to reach your audience quicker. It’s not really a bad idea to provide MVP with this approach and go native if you see traction. Plus you reach all and not just Android or iOS users.</li>
<li>**Enterprise Apps. **If you are a company and want to promote BYOD, the Native approach can get expensive if you try making it work for all brands, versions, and different kinds of mobiles unless you force your employees to use one kind of Phone. 😉</li>
</ol>
There are various great Hybrid app development platforms now. Here are a few which are trending:
a. Ionic
b. Titanium
c. Sencha Touch
d. PhoneGap
<ul>
<li><a href="https://medium.com/tag/mobile-app-development?source=post">Mobile App Development</a></li>
<li><a href="https://medium.com/tag/native-vs-hybrid?source=post">Native Vs Hybrid</a></li>
<li><a href="https://medium.com/tag/startup?source=post">Startup</a></li>
</ul>
Credit: Mitesh Bulsara, Serial Entrepreneur and Coach, CTO, CoffeeBeans

This is an old and dusted topic the points I intend to discuss here are based on my 7 years of experience with mobile app development.

Hybrid vs Native: Old Battle — New Approach

Asha, one of the finest employees in her organization, fell into a financial crisis a year back, for reasons outside her control. She missed some payments on EMI and her credit score suffered. This is making it difficult for her to get a new job and is worsening her financial situation. Which in turn, is making it even more difficult for her to find a job.
This isn’t the truth, at least as yet. But you know that you are not very far from this reality. The reality where algorithms will start judging you and take more decisions for you. And before you know it, you will be in an Algorithmic Prison.
<img src="https://media.coffeebeans.io/blogs/Algorithm-Prison-1-2048x1152.jpg" alt="cover">
We are increasingly seeing the use of Machine learning algorithms in predicting multiple facets of life. Predicting if it is going to rain tomorrow is one thing, predicting whether a loan applicant will default, is an entirely different proposition. The algorithms are deciding if you should get the loan, if you are fit for a job, or if you are eligible for that insurance. Although, for companies, the loss of missing out on opportunities on risky propositions may not be worth it; for individuals, it could be life-changing and have very high negative repercussions.
To make matters worse, through connected systems, if you are denied one service, it is likely you will be deemed unfit for multiple such opportunities.
The Bias
The algorithms study multiple data points and behavioral aspects of the past and give out probabilistic predictions. If the past was biased, it is very likely the predictions will show that too. Thereby, possibly depriving certain regions and castes, and religions of opportunities. While we morally preach equality for all, the algorithms do not have moral values. At least not as of now.
Moreover, the algorithms make predictions while segmenting you into buckets. If on average, people like you from the same region, company, or race have defaulted, it doesn’t mean you are going to default too
The Polarised Views
If you are consuming a kind of content. the engines show similar content to you in anticipation of increasing click-throughs. Thereby, just adding to the confirmation bias of your opinions and creating polarised views. We are already seeing the impact of social media on election campaigns. And then there are companies, who are trying hard to use those analytics and help political parties. Media is the only source of information for you, and that is highly driven by algorithms on what to show you next. The movie Inception… its already here, just not in the form of the movie.
The Justice
Most of the algorithms often keep learning with data and make decisions on their own. This makes the systems highly opaque and difficult for a human to reason why a particular decision was made.
We have a proper judiciary system in place. But if an algorithm is denying you a job or a loan, who is to be blamed for that? If it is discriminating against a race or religion, can you explain how the algorithm made the decision? Well, the justice system has to evolve. For all you know, that will also be driven by algorithms.
Privacy — Huh?
You have thought of this at least once that Alexas, Siris, and Googles of the world are “listening”, to you. We have no clue how the companies are tracking our behavioral data for the purpose of providing the best recommendations and services possible.
Monika, a friend, through her privacy awareness, has enabled the highest security an app has to provide. But, there are services that are part of our lives now. Choosing complete privacy is not possible now. We all have to be socially connected. She has to use Maps frequently and share her location. Privacy — well, yeah!!!
Do the right thing
AI is the new hype and we are seeing new prediction/recommendation engines emerging every day. All the systems are being built for a reason and the reasons may vary from being as simple as generating more profits, to as serious as the security of a Nation. It is changing our world entirely, but the consequences may reach far beyond the original purpose.
We understand most of the data is about people’s behaviors and identity. We value and respect it with humanity, while consciously taking responsibility for the world that we want to live in.
An algorithm or software is not good or bad in itself. It is important how we use it. It is up to us, the engineers and the organizations in this domain, to make ethical choices, take responsibility, and Do the right thing
1)Machine learning
2)AI
3)Data Science

Asha, one of the finest employees in her organization, fell into a financial crisis a year back, for reasons outside her control. She missed some payments on EMI and her credit score suffered. This is making it difficult for her to get a new job and is worsening her financial situation. Which in turn, is making it even more difficult for her to find a job.

You are in an Algorithmic Prison. And the funny thing is, you don’t even know what it is

So you have an idea? Great. Many people are going to tell you to quickly come up with a prototype or an MVP and validate the acceptance of the idea. You start working on it and a month goes by for your MVP to shape up. Awesome.
What if I tell you that you can start validating your idea the very day the idea strikes you? **Have you heard of Dropbox getting famous without any servers but just with a video? **Well, with my experience I can say that there are many cases when you can confirm your idea without the MVP itself. That’s how I did it.
<img src="https://media.coffeebeans.io/blogs/Get-leaner-1-2048x1152.jpg" alt="cover">Motorpsychos, is a motorcycle renting service in Pune, India. When this idea struck me, there was a dire need for such a service in Pune.
The most deceiving thing about any business idea is that — they all sound great.
Then I made a move, which turned out to be a smart one. Instead of going with the MVP and buying a second-hand motorcycle to see how many users are actually renting, I did something else. It took me one day to create a quick website with a contact-us page. I uploaded information about the service on a few blogs. I registered myself on Justdial, Yellow Pages as a Motorcycle renting agency. I didn’t procure any vehicle.
I just wanted to see if anyone would respond to Motorpsycho’s. To my surprise, on the very 2nd day, I got my first call. But then for a week, nothing happened (I am guessing the 2nd-day call was from a competitor). Suddenly, after one week I started getting 10 to 11 calls a day requesting motorcycles on rent.
The point is if people need something they will find you out. You just have to be easily available.
Way to go. Demand and Market were validated. But I still had doubts about what motorcycles to buy? How much the rent should be? Here are the 5 smart things that I learned without an MVP.
<ol>
<li>I didn’t have any bike to cater to. So when people asked for it, rather than telling them that we don’t have bikes available, I would say “Sorry, we are overbooked and have waitlisted you.” This gave them a feeling that we are doing good and are trusted by others who are renting from me.</li>
<li>I also got an idea of what motorcycles they are interested in renting. If they would ask for a 150 cc vehicle, I would say “It’s not available but I can check the availability of 200 ccs and call you back if you are interested” just to see if they are okay with that. This helped me immensely in decision-making while procuring vehicles.</li>
<li>Before starting the business, I got a fair idea of how much the customer is willing to pay as rent. This gave me a good heads-up with regard to my turnover before I procured a single bike.</li>
<li>I got an idea of what segments of people to target. All the while I was guessing that college-goers will be my major target audience. But to my surprise, there were many working girls too calling for 2-wheelers. I was able to concentrate and focus on my marketing in a better way.</li>
<li>Motorpsychos was already a brand before the business started. This approach wouldn’t have been successful if I had not been extra nice and extra polite to the users who called. I was treating them personally. I followed up with them even if I couldn’t rent them a vehicle, saying “Sorry, you were wait-listed but there were no cancellations, etc.”</li>
</ol>
Having MVP is great, but not always necessary. Remember, the customers always want the moon if you ask them before starting a business. But most of the time, they just want someone to make them feel special and then they will buy whatever you sell.
<ul>
<li>
<a href="https://medium.com/tag/entrepreneurship?source=post">Entrepreneurship</a>
</li>
<li>
<a href="https://medium.com/tag/growth-hacking?source=post">Growth Hacking</a>
</li>
<li>
<a href="https://medium.com/tag/startup?source=post">Startup</a>
Credits: MItesh Bulsara, Serial Entrepreneur and Coach, Chief Technology Officer, CoffeeBeans consulting
</li>
</ul>

So you have an idea? Great. Many people are going to tell you to quickly come up with a prototype or an MVP and validate the acceptance of the idea. You start working on it and a month goes by for your MVP to shape up. Awesome.  What if I tell you that you can start validating your idea the very day the idea strikes you? **Have you heard of Dropbox getting famous without any servers but just with a video? **Well, with my experience I can say that there are many cases when you can confirm your idea without the MVP itself. That’s how I did it.

Get leaner. Don’t waste your time on MVP to validate your idea

Proof of concept (PoC) vs prototyping vs spike!!!!
Frankly to me, all are same and throughout this article, I will stick to the term PoC.<img src="https://media.coffeebeans.io/blogs/Art-of-PoC-2048x1152.jpg" alt="cover">Looking up for definition of PoC I stumbled upon wikipedia which says,
“ Proof of concept (PoC) is a realization of a certain method or idea in order to demonstrate its feasibility, or a demonstration in principle with the aim of verifying that some concept or theory has practical potential.”
Phew!!! That’s a very heavy definition.
To me, PoC is something that proves if stuff works or not. Anything beyond this and the whole fun around PoC gets sucked out.
A new project is about start and it comes with its own package of unknowns. 3rd party integrations? Explore a new graph-based db? Flink vs Spark?
How does one estimate these unknowns? Obviously one can spend an eternity answering these questions but clients really don’t care about your anxieties.
They just care about time. And making blind assumptions without any validation can make or break a project.
That’s where PoCs are invaluable to quickly and efficiently validate our assumptions and gain clarity in regard to making important decisions.
<ul>
<li>Timebox — Based on my experience, if a PoC extends beyond one week (5 working days), then either we are trying to prove too many things as part of this one PoC or the thing we are trying to prove is….not PoC worthy and deserves its own project timeline. PoCs are supposed to be quick and very laser-focused. One week is my rule of thumb.</li>
<li>Document — This aspect is something the developer is oblivious to when he is busy writing code and debugging stuff at 3X speed. But once the PoC is complete and everyone is happy, the learnings are almost every time lost and nobody cares. These learnings can apply to multiple projects and save a lot of time for others and decent documentation (not a very formatted and extensive one) which covers all the learnings is worth saving. It’s not just enough to document the learnings but the hacks and tricks you needed to make for the PoC to work like editing some configuration parameters, tweaking some environment variables, etc. The online resources you have referred to are something worth documenting. I can’t stress enough on this point but documentation is a must.</li>
<li>Write messy working code but not that messy — At least in most of the projects I was involved in, the code used in PoC usually ends up being reused as the base code for the actual project. So it actually makes sense to write some comments and have atleast a basic structure and good naming conventions which can allow others to read and debug the code. I am okay with one gigantic main method but please don’t make the code unreadable with variable names like x1, t2, etc.</li>
<li>Commit the code — Once a PoC is complete it contains not just one java class file but multiple other files like scripts, build files, docker files, etc. Its important to commit all these files in a code repo instead of keeping it on our personal laptop and when asked for the code after 6 months, doing crazy searches across all your drives because as devs we usually end up with file naming conventions like Test1.java, file1.sh, etc.</li>
<li>Demo — Now that you have completed the PoC and discussed the findings with your architect which helped him conclude, you feel satisfied and happy. Want to feel proud, then share the findings with the team as part of a demo. Showcase the whole working thing to the team in a live session and enjoy as others gasp at the magic you have created in just a few days. But it’s not just about you feeling proud. It’s making sure others validate the approach you took and come up with some smart/creative ideas which can take your code to the next level. A discussion can generate many cool ideas as opposed to others reviewing your document in silos.</li>
</ul>
Hope whatever I have covered makes sense and would love to hear others’ thoughts.
Also, please don’t forget to have fun. At the end of the day its awesome to see something running and feel a few points smarter.
<h5>Credits: Sarthak Kumar Barik, Solution Architect</h5>

Looking up for definition of PoC I stumbled upon wikipedia which says, “ Proof of concept (PoC) is a realization of a certain method or idea in order to demonstrate its feasibility, or a demonstration in principle with the aim of verifying that some concept or theory has practical potential.

Art of PoC

You may have heard of DevOps which revolutionized the software development industry. MLOps where ML stands for Machine Learning leverages AI to deliver superior performance as well as value. It includes some of the core concepts associated with DevOps. As per industry experts, the worldwide MLOps market will amount to US $75.42 Billion by the year 2033. In this article, we will learn what MLOps is and the trends and predictions regarding this cutting-edge technology.
<h2><img src="https://media.coffeebeans.io/The%20Future%20of%20MLOps.jpg" alt="cover">What is MLOps?</h2>
MLOps is a set of practices with the end goal of deploying as well as maintaining ML models in production in a stable as well as efficient way. Other relevant goals include streamlining, monitoring as well as automating the relevant machine-learning models.
The current market is experiencing rising demand for optimized and flexible MLOps tools. MLOPs leverage AI as well as borrows principles belonging to DevOps. As a result, machine learning workflows are augmented with typical DevOps properties such as continuous integration as well as collaboration.
We take a look at the top emerging trends of the MLOps Platform.
<ul>
<li>Mounting on Edge AI as well as On-Device</li>
</ul>
Computation as well as data storage devices are positioned in relative proximity to the source. To your knowledge, users are inclined towards edge deployment. The reasons include ramped-up data as well as client privacy. Machine learning solutions experience more responsiveness, value addition and stability.
<ul>
<li>Merging with DevOps as well as DataOps</li>
</ul>
This trend enhances collaboration as well as optimizes processes across existing teams. Incorporating DataOps results in good data quality as well as optimized management of data pipelines. Including DevOps facilitates continuous integration as well as delivery pipelines.
<ul>
<li>The Process will become increasingly automated with time and evolvement</li>
</ul>
Automation will reduce dependence on humans and the quantity of errors. Performance and efficiency will be among the areas where users of MLOps will experience growth. Particularly tasks such as deployment of models as well as tracking will likely be automated. Automation will increase the value and benefits of MLOps in myriad ways.
<ul>
<li>Model Interpretability as well as Explainability</li>
</ul>
There is a rising requirement of interpretable as well as transparent machine learning models. The technology emphasizes including tools as well as practices for why model decisions took place. Lastly, there will be the need for necessary regulatory compliance.
<ul>
<li>Rising Emphasis on AI Ethics and Governance</li>
</ul>
As many of us realize, AI technology cannot be left to determine all decisions. Governments and regulatory bodies across the world will enact rules to ensure ethics are not violated, fairness is maintained and there is sufficient transparency in the implementation of MLOps.
<h3> Here are the top predictions for MLOps in 2024</h3>
<ul>
<li>Diverse and a greater number of industries will utilize MLOps</li>
</ul>
With rapidly growing awareness about the relevance as well as usefulness of MLOps many firms spanning different verticals will adopt MLOps in their day-to-day operations. Currently numerous industries including retails, healthcare as well as financial services are implementing MLOps. The latter enables novel drug discovery as well as precise diagnosis and tailored treatment plans. The retail sector gains valuable insights such as superior demand forecasting and adept customization. The Finance industry gets assistance in efficient as well as effective fraud detection. Other benefits include enhanced risk management as well as optimized trading.
<ul>
<li>Universal Availability of this State-of-the-art technology</li>
</ul>
With the passage of time, MLOps will expand to more and more areas and audiences. Company staff such as domain experts, programmers as well as data scientists will actively participate in leveraging as well as implementing this exciting and futuristic strategy. We call this phenomenon the democratization of MLOps.
<ul>
<li>Improvement and Refinement of MLOps</li>
</ul>
This domain will only improve with time and data inputs. Training on newer and more complex data will result in the improvement of MLOps. It will be more up-to-date, precise as well as effective. New versions of MLOps will emerge in the market and it will be a highly sought-after strategy. There will be increasing demand for human resources well versed with MLOps. Experts with knowledge of MLOps will be offered higher salaries and better positions.
<ul>
<li>Maturing and Acceptance of MLOps</li>
</ul>
MLOps has enormous potential to grow to new heights and expand in other areas. The process will only mature and improve with time. The cost of learning and implementing MLOps will decrease resulting in universal adoption of this AI-powered process. Customers and Clients will demand the leveraging of MLOps by companies. Companies utilizing MLOps will have a competitive edge in the market.
<ul>
<li>Rise in Cloud-Native MLOps </li>
</ul>
As a result of the popularity of Cloud-native MLOps, entities will get access to scalable as well as flexible tools. The platforms are designed keeping MLOp's needs and processes in mind. One can deal with model development as well as deployment while not having the responsibility of the management of the underlying infrastructure.
<a href="https://coffeebeans.io/">CoffeeBeans</a> has the requisite expertise as well as relevant experience in implementing MLOps. Over the years we have built up an impressive track record in this important and indispensable domain. Contact us to know firsthand how we can facilitate your organization's seamless transition to MLOps. Make sure you do not lag behind your competitors by implementing top-notch and efficient MLOps solutions in your area of expertise.

You may have heard of DevOps which revolutionized the software development industry. MLOps where ML stands for Machine Learning leverages AI to deliver superior performance as well as value. It includes some of the core concepts associated with DevOps.

The Future of MLOps: Trends and Predictions Shaping the Next Generation of ML Operations

Apache Kafka, the open-source, distributed event streaming platform, has been a game-changer in the world of data processing and integration. It has been continuously evolving over the years, offering new features and improvements that make it more scalable, secure, and efficient. The latest release of Apache Kafka, version 3.3, brings a number of exciting changes, including the replacement of Zookeeper and the introduction of a new consensus mechanism called KRaft.
<img src="https://media.coffeebeans.io/blogs/1_SxN1MbaEhoRsIGImLTLs-A.jpg" alt="cover">
The Evolution of Apache Kafka’s Consensus Mechanism
One of the main changes in Apache Kafka 3.3 is the replacement of Zookeeper with a new consensus mechanism called KRaft. Zookeeper has been a critical component of Apache Kafka for many years, providing a centralized configuration service for the platform. However, Zookeeper has been a bottleneck for large-scale, high-throughput deployments, leading to limitations in scalability and reliability.
KRaft is a distributed consensus algorithm that provides strong consistency guarantees while also being highly scalable. Unlike Zookeeper, which operates in a leader-based fashion, KRaft operates in a leaderless fashion, meaning there is no central point of failure. This provides faster and more efficient leader elections and lower latencies, making it a more reliable and scalable solution for Apache Kafka.
Security Enhancements in Apache Kafka 3.3
In addition to the replacement of Zookeeper, Apache Kafka 3.3 also brings a number of security enhancements. This includes support for Mutual TLS (Transport Layer Security), which provides an additional layer of encryption for communication between Kafka clients and brokers. This enhances the overall security of the platform, making it a better choice for sensitive and regulated environments.
Apache Kafka 3.3 also includes improved support for Kerberos authentication, which is a widely used security protocol for secure communication over networks. These enhancements make Apache Kafka even more secure, ensuring that data transmitted through the platform is protected against unauthorized access and tampering.
API Enhancements in Apache Kafka 3.3
Apache Kafka 3.3 brings a number of API enhancements that make it easier for developers to build and integrate the platform into their applications. This includes new APIs for managing topics and partitions, as well as improvements to the existing APIs for producing and consuming messages. These enhancements make the platform more user-friendly and accessible, providing developers with more options for customizing and integrating Apache Kafka into their applications.
The Power of Apache Kafka Connector APIs
Another major feature of Apache Kafka 3.3 is the addition of new connector APIs. These APIs make it easier to integrate Apache Kafka with other systems and data sources, such as databases and data lakes. This opens up a whole new world of possibilities for data ingestion, processing, and analysis, making Apache Kafka an even more powerful tool for building modern data-driven applications.
Conclusion
In conclusion, Apache Kafka 3.3 is a major milestone for the platform, offering a number of exciting new features and enhancements that make it even more scalable, secure, and efficient. From the replacement of Zookeeper with KRaft, to the new security and API enhancements, this release makes Apache Kafka an even better choice for event streaming and data integration. Whether you’re a developer looking to build a new application, or an IT professional looking to modernize your data architecture, Apache Kafka 3.3 is the perfect solution for your needs.
Keywords: Apache Kafka, Zookeeper, KRaft, event streaming, data integration, security, API enhancements, connector APIs, Mutual TLS, Kerberos authentication, distributed consensus algorithm.\
Credits: Greeshma Pillai, <a href="https://coffeebeans.io/">coffeebeans</a>

Apache Kafka, the open-source, distributed event streaming platform, has been a game-changer in the world of data processing and integration. It has been continuously evolving over the years, offering new features and improvements that make it more scalable, secure, and efficient. The latest release of Apache Kafka, version 3.3, brings a number of exciting changes, including the replacement of Zookeeper and the introduction of a new consensus mechanism called KRaft.

Discovering the Innovations in Apache Kafka 3.3: From Zookeeper to KRaft

An embedded database is a database management system that is integrated within an application rather than being a standalone system. This means that the application has direct control over the database, and the database runs within the same process as the application. Embedded databases are often used in applications where data needs to be stored and accessed locally, without the need for a separate database server.<img src="https://media.coffeebeans.io/blogs/1_iLk9R-LEmBjLw4cVYaehtw.jpeg" alt="cover">Embedded database
Embedded databases are typically used when an application needs to store data locally and the data is not shared with other applications or users. They are often used in desktop and mobile applications, as well as in embedded systems such as IoT devices, where network connectivity may be limited or unreliable.
Some common examples of embedded databases include SQLite, H2 Database, Berkeley DB, and Firebird. These databases are often lightweight, fast, and efficient, making them ideal for use in applications with limited resources. They may also include features such as ACID transactions, replication, or full-text search, depending on the specific needs of the application.
When to use one?
Embedded databases are typically used when an application needs to store data locally and the data is not shared with other applications or users. They are often used in desktop and mobile applications, as well as in embedded systems such as IoT devices, where network connectivity may be limited or unreliable.
One popular embedded database that deserves special mention is Watermelon DB. Watermelon DB is an open-source, reactive database that is designed specifically for use in client-side applications. It provides a fast and reliable way to store data locally, with built-in support for offline data synchronization.
Watermelon DB is ideal for use in mobile and web applications that need to work offline. It uses a data synchronization protocol called Differential Synchronization, which allows the database to synchronize with a remote server while minimizing bandwidth usage and reducing conflicts. Watermelon DB is also highly scalable, making it suitable for use in applications with large and complex data models.
Some common use cases for embedded databases include:
● Desktop and mobile applications that need to store data locally.
● IoT devices that need to store sensor data locally.
● Point-of-sale systems that need to store transaction data locally.
● Gaming applications that need to store game data locally.
● Medical devices that need to store patient data locally.
Popular Embedded databases:
● SQLite: A lightweight, open-source database that is widely used in mobile and desktop applications.
● H2 Database: A Java-based database that is designed to be fast and efficient.
● Berkley DB: A high-performance, embeddable database that is used in a wide range of applications.
● Firebird: An open-source relational database that is used in embedded systems and desktop applications.
<img src="https://media.coffeebeans.io/blogs/1_7tGLxM6IBmX2VaJnIb1iwg.png" alt="database">Embedded databases and its feature
How to choose the correct embedded database?
Choosing the correct embedded database for your application depends on a number of factors, such as the size and complexity of the application, the performance requirements, and the desired features. Here are some steps you can follow to choose the right embedded database for your needs:
· Identify the requirements of your application: Determine what kind of data needs to be stored and how it will be accessed, as well as any performance or scalability requirements.
· Consider the size of the database: Depending on the size of your application and the amount of data you need to store, you may need to choose a database that is lightweight and efficient. Look for databases that have a small footprint and use minimal system resources.
· Evaluate the performance: Consider the performance characteristics of the database, such as query times and transaction processing speeds, to ensure that it meets your requirements.
· Look for required features: Determine which features are essential for your application, such as support for ACID transactions, replication, or full-text search. Make sure the database you choose includes these features.
· Check the licensing: Some embedded databases have more restrictive licenses than others. Make sure the license of the database you choose is compatible with your application’s requirements and licensing model.
Embedded databases are an important component in many desktop and mobile applications, as well as in embedded systems such as IoT devices. These databases provide a lightweight and efficient way to store data locally without the need for a separate database server.
At CoffeeBeans, we have experience and expertise in using embedded databases to build reliable and efficient solutions for our clients’ offline data storage and synchronization needs. If you’re looking for a partner to help you implement offline capabilities in your mobile or web application, we can help. Our team of experts can help you design and implement a solution that meets your needs and provides a seamless offline user

An embedded database is a database management system that is integrated within an application rather than being a standalone system. This means that the application has direct control over the database, and the database runs within the same process as the application. Embedded databases are often used in applications where data needs to be stored and accessed locally, without the need for a separate database server.

Embedded Databases Decoded: Choosing the Right Type for Your Application’s Success

DeFi, short for Decentralized Finance, is a movement that aims to offer financial services and products that are open to everyone, without the need for intermediaries. DeFi has taken the world of cryptocurrency by storm and is quickly becoming one of the hottest trends in the industry.
<img src="https://media.coffeebeans.io/blogs/Defi.png" alt="cover">One of the key features of DeFi is that it operates on a decentralized platform, meaning that it operates on a network of computers rather than relying on a central authority such as a bank. This makes DeFi transactions faster, cheaper, and more secure compared to traditional financial transactions.
Financial services performed on public blockchains, particularly Ethereum, are referred to as Defi (or “decentralized finance”). With DeFi, you can earn interest, borrow, lend, buy insurance, trade derivatives, swap assets, and do other things, but it’s faster and doesn’t require paperwork or a third party. DeFi, like crypto in general, is worldwide, peer-to-peer (meaning it is sent directly between two people rather than being routed through a centralized system), pseudonymous, and open to everybody.
How Blockchain Plays a Critical Role in DeFi
Blockchain technology is a critical component of DeFi as it enables trust less transactions, eliminates intermediaries, and ensures data immutability. With the use of smart contracts, DeFi eliminates the need for intermediaries and ensures that transactions are executed automatically, transparently, and securely.
Cryptocurrencies leverage blockchain technology, which is employed in decentralized finance. A blockchain is a decentralized and secure database or ledger. The blockchain is managed and transactions are processed via applications known as dApps.
The blockchain stores transactions in the form of blocks, which are later confirmed by other users. If all of these verifiers agree on a transaction, the block is closed and encrypted, and a new block is created with the previous block’s details within.
The name “blockchain” refers to how the data in each succeeding block “chains” the blocks together.
One popular DeFi application is Defi Swap, which allows users to exchange different cryptocurrencies without having to go through a centralized exchange. Defi Swap operates on smart contracts, which are self-executing contracts with the terms of the agreement between buyer and seller being directly written into lines of code. This makes Defi Swap transactions completely transparent and secure, as the terms cannot be altered once the contract has been executed.
Another exciting aspect of DeFi is the use of DeFi Coins. DEFC, for instance, is a DeFi Coin that provides access to a wide range of DeFi services, including Defi Swap. DEFC is unique in that it operates on a dual-chain platform, which means that it operates on both Ethereum and Binance Smart Chain. This allows DEFC to offer fast, cheap, and secure transactions, making it an attractive option for DeFi users.
Defi Advantages and Risks
Defi Advantages
· No Boundaries– DeFi has no territorial restrictions, as is the case with everything in the realm of cryptocurrencies. Theoretically, anyone with access to the Internet and cryptocurrency wallets could communicate with DeFi.
· Greater Speed– A lack of middlemen and borders, along with blockchain technology’s greater speed, leads to quicker transactions and generally simpler access to the services.
· Greater Transparency– A key benefit of various blockchain technologies is the total openness that exists between all stakeholders.
· Programmability– Smart contracts’ flexibility can better-automated automate solutions, which in turn can be utilised to produce new assets and financial instruments.
· Interoperability– The system’s flexibility also makes it simpler to create centralized and decentralised applications (dApps) that complement one another at a deeper level than is conceivable in the realm of conventional financial solutions and services.
· Immutability– The immutability of a blockchain-based decentralised system based on cryptography and data protection promotes security and trust.
· More Command– A single user has more control over their finances thanks to DeFi’s permissionless, transparent, and clear decentralised architecture. To keep track of their DeFi statistics and portfolio, there are numerous alternatives.
Defi Risks:
· Regulatory Danger– DeFi protocols are now operating with absolutely no government control or regulation from any government organisation. Simply said, this position is subject to change, and it is impossible to forecast how new government legislation governing DeFi protocols may influence your DeFi investments.
· Token Danger- Each DeFi investment includes cryptocurrency tokens. For example, if you deposit cash in a Uniswap liquidity pool, your investment exposes you to the pool’s two tokens: the liquidity provider token that you receive after depositing and the Uniswap token that you earn as a reward. If you invest in a stablecoin pool, the pool will most likely contain a mix of different stablecoins.
· Consumer Protection is Lacking- DeFi has thrived in the absence of laws and regulations. However, when things go wrong, users usually have little or no protection. There are no state-run DeFi compensation mechanisms, and no restrictions requiring DeFi service providers to maintain capital reserves.
Here Are Some of The Most Impactful Defi Uses:
1. Decentralized exchanges (DEX): A decentralized exchange is a platform that enables peer-to-peer trading of cryptocurrencies without the need for a central authority. DEXs use smart contracts to automate the process and ensure transparency and security.
2. Lending and borrowing: DeFi lending and borrowing platforms allow users to lend and borrow cryptocurrencies, often with adjustable interest rates. These platforms use smart contracts to automate the lending process and ensure security and transparency.
3. Stablecoins: Stablecoins are cryptocurrencies that are pegged to a stable asset, such as the US dollar, to minimize price volatility. They play a critical role in DeFi by providing a stable medium of exchange for decentralized financial transactions.
4. Yield farming: Yield farming is a practice of lending cryptocurrencies to DeFi protocols to earn interest. It allows users to earn passive income by participating in the liquidity of DeFi protocols.
5. Insurance: DeFi insurance protocols offer insurance coverage for decentralized financial transactions. They use smart contracts to automate the claims process and ensure security and transparency.
6. Predictive markets: Predictive markets are decentralized platforms that allow users to trade on the outcome of events, such as elections or sports games. They use smart contracts to automate the settlement process and ensure transparency and security.
7.Crowdfunding: DeFi crowdfunding platforms allow users to invest in projects, often in exchange for a token that represents a stake in the project. They use smart contracts to automate the investment process and ensure security and transparency.
8. Identity management: DeFi identity management protocols allow users to store and manage their personal information in a decentralized and secure manner. They use blockchain technology to ensure that users have control over their data and that it cannot be tampered with.
9. Asset tokenization: DeFi asset tokenization protocols allow users to convert real-world assets, such as real estate or art, into tokens that can be traded on blockchain networks. They use smart contracts to automate the process and ensure security and transparency.
These use cases demonstrate the diverse range of financial services that DeFi protocols can provide, and they are just a few examples of how DeFi is changing the way we think about and use money
Defi Predictions for Future
The DeFi Coin token price is now sitting at $0.3578, following some impressive upward momentum in recent weeks. Let’s take a closer look at the coin’s long-term prospects, so you know what kind of profits you may expect from a DEFC investment:
End of 2023: The DeFi sector’s development trend is showing no signs of slowing down, which is expected to entice investors away from established financial systems. If this happens, according to our DEFC price projection, the currency may be worth $0.95 by the end of 2023.
End of 2025: In the coming years, DeFi Coin and DeFi Swap will probably keep growing inside the DeFi market, providing a growing number of opportunities for investors to earn returns on their holdings. We predict that the price of DeFi Coin may reach $1.40 by the end of 2025 since the price of DeFi Coin should logically increase along with this expansion.
DeFi is transforming the world of finance and is at the forefront of the decentralized finance revolution. With its secure, transparent and decentralized platform, DeFi has the potential to provide financial services to millions of people who are currently excluded from traditional financial systems. Although DeFi is still in its early stages, the future looks bright and we can expect to see more and more people adopt.
Credits : Greeshma Pillai, CoffeeBeans

DeFi, short for Decentralized Finance, is a movement that aims to offer financial services and products that are open to everyone, without the need for intermediaries. DeFi has taken the world of cryptocurrency by storm and is quickly becoming one of the hottest trends in the industry.

Decentralized finance (DeFi) and its future

From sourcing raw materials from every corner of the world and selling end products across the world, Supply chain management plays a vital role! And technology never failed to do its job! Blockchain has already disrupted the finance industry and the supply chain will be the next
<ul>
<li>By Siva Seetharaman, <a href="https://coffeebeans.io/">CoffeeBeans</a> Consulting</li>
</ul>
<img src="https://media.coffeebeans.io/blogs/1_bIG3LRY-CNqs0qSAUXqvFw.jpg" alt="cover">
Supply Chain Management is becoming one of the most critical areas for businesses to thrive. Manufacturing companies source raw materials from every corner of the world and finished goods are also moved across borders. As the quantum of imports/exports increases, there is a risk that SCM might become a bottleneck especially if there is a lack of planning.
Technology has indeed played a crucial role in transforming this age-old sector of movement of goods. Large enterprises are deploying sophisticated Enterprise Resource Planning systems, and Transport Management Systems to manage their day-to-day operations. But still the current processes and the systems do have their own deficiencies. These deficiencies reduce the credibility of the whole SCM infrastructure.
Challenges in current Supply Chain Infrastructure:
<ol>
<li>As goods get transported from one corner to another corner of the world, one shipment is handled by multiple stakeholders until it reaches the customer. Practically, there cannot be one single ERP that is implemented by all the players in the chain.</li>
<li>The lack of one system results in lack of transparency in the flow of Inventory and Information among the participants. The participants operate in silos and parties like banks and end customers are also often kept in the dark.</li>
<li>As the Supply Chains involve a lot of cross-border transactions, the possibility of mixing counterfeit products into the chain is very much possible and current systems are not sophisticated to identify this and trace it back to the source.</li>
<li>There are a lot of manual data entries and paperwork to be done in the current setup. This compounds missing information, and erroneous transactions resulting in reputational and monetary losses in many cases.</li>
</ol>
Blockchain Technology:
Blockchain Technology is now the buzzword in the world of technology. Blockchain has already made a huge impact in the finance industry. Although there are proponents and opponents to Bitcoins and other currency tokens, the idea of blockchain that makes this happen is indeed foolproof and niche. Following the Finance sector, other sectors have started to explore possibilities to find answers to their problems from Blockchain.
What is Blockchain?
Blockchain is a decentralized digital ledger system that has the capability to receive and record transactions from multiple parties involved in the system. In a traditional database, data is stored in the form of tables. But in a blockchain, the data is grouped and stored as blocks which are in turn connected to existing blocks thus forming a blockchain. A block is the storage unit in the blockchain where transaction information is stored. Along with transaction information, other details like an encrypted number from the previous block, block header, etc. are also stored. Building the blocks involves multiple people vouching for the credibility of the transactions that happen between two parties. Only after transactions are validated, a block will get added to the existing blockchain. It is nearly impossible as of now to change or delete data that goes into the blockchain. The unique feature of blockchain when compared to the traditional database systems is that the data that enter the blockchain are tamper-proof and are recorded forever.
How can Blockchain disrupt the Supply Chain Industry?
As mentioned earlier, blocks are the basics of a Blockchain. Out of the many advantages of this system, three points stand out clearly as game-changers when it comes to implementation in Supply Chain Management. They are Improved Traceability, Minimized human intervention, and Enhanced coordination between the parties. Before getting into the advantages of a Blockchain system in Supply Chain, it is important to understand how to build a blockchain system in a supply chain scenario.
A Classic Supply Chain scenario:
<img src="https://media.coffeebeans.io/blogs/0_RMLMhIkn0M8JHU3y%20(1).png" alt="supply chain">
In the above scenario,
<ol>
<li>A customer places an order with a retailer</li>
<li>To fulfill the requirement, the retailer in turn orders to the distributor</li>
<li>The distributor then places the order to a manufacturer to produce the product</li>
<li>Once the manufacturing is done, the manufacturer fulfills the distributor. The distributor fulfills the retailer</li>
<li>As a final step, the retailer fulfills the customer order.</li>
<li>The retailer or distributor or manufacturer might seek the support of a bank for financial aid.</li>
</ol>
As mentioned above, blockchains can improve the credibility of the system by improving the transparency of flow of information and the traceability of inventory between the parties. But why should a system be transparent? Let’s read the cases below to understand why.
Case 1: Banks should make informed decisions before approving a loan or any other transactions. Let’s assume a case where the manufacturer applies for a loan in the bank. In this case, the bank will have to conduct physical audits of the transactions and documentation before approval. This is a long and error-prone process. In a best-case scenario, if the bank can access all the upstream transactions along with agreements in one place, then it will be easy and foolproof for the bank to approve such requests.
Case 2: Let’s assume the distributor imports finished goods from a manufacturer who is in a different country. At the time of receiving at the port, some counterfeit goods are mixed in the supply chain. When the distributor finds this, he will not be able to figure out who mixed counterfeit into the chain. The recall and reverse logistics will then stand out as new problem statements. To address this problem, the distributor should be part of a system where every single transaction with every single participant is recorded in a single system.
A Blockchain Supply Chain system:
<img src="https://media.coffeebeans.io/blogs/0_9exw-HhJUf-7PdjU.png" alt="supply chain system">
For a blockchain system to work as expected, all the parties who are participating in the supply chain should agree to be part of the blockchain. In this case, the assumption is all the parties mentioned in the above diagram are part of the same blockchain. (Reference <a href="https://hbr.org/2020/05/building-a-transparent-supply-chain">https://hbr.org/2020/05/building-a-transparent-supply-chain</a> )
<ul>
<li>The colored bars mentioned in this diagram are the blocks that get created whenever a transaction is done between two participants.</li>
<li>In the first step, the customer orders a product from the retailer and the retailer gives back an order confirmation. These are two separate transactions. So two separate blocks are created. The green block is the customer side and the blue block is the retailer side transaction.</li>
<li>As the flow moves forward, the blocks keep added to the blockchain. So from customer ordering to customer fulfillment, all the blocks are part of this blockchain and anyone participating in this chain will gain the full visibility of the same.</li>
<li>These blocks now form the blockchain. Every transaction is recorded and every participant of the chain can now access the data. This makes the system completely foolproof and also establishes the ultimate transparency.</li>
</ul>
Now that we know how a blockchain system works in a Supply Chain scenario, below are the key aspects of the technology that will transform the sector.
Improved Traceability: Every step of the flow is entrusted and recorded as blocks. This makes it easier for the parties to trace back shipments to the source. This is highly useful in sectors like farming, and fishing where multiple unknown parties are involved at source. Also, in sectors like Pharma, Art, it is possible to trace back and figure out the exact point when and where a counterfeit came into the supply chain.
Minimized human intervention: Airway bills, Invoices, Handling instructions, Bill of Landing, Export License, and Packing list are some of the many documents that are involved in a supply chain. The worrying part is that the maker checker process in documentation is purely manual and it is prone to errors. With Smart Contracts in Blockchain, the complete industry can go paperless and improve the efficiency multifold.
Enhanced coordination between the parties: If all the parties agree to get on a blockchain, then there will be a seamless flow of information to all the participants. This improves the coordination between the parties in the supply chain and the overall visibility of information and inventory flows.
Conclusion:
The implementation of Blockchain in Supply Chain Management has been tested in the real world by a few companies like Walmart Canada, IBM Food Trust. But these are tested on a small scale and the world is yet to see a full-fledged implementation where Blockchain replaces ERP.
Transparency of information flow between parties, Reduction of IT costs, Elimination of manual efforts, and Reduction of losses incurred due to counterfeit products are some of the key advantages of Blockchain technology over traditional technology. There are different ways of implementing a blockchain network. The network can be either public or private. Also, there are various other value adds like smart contracts which reduce the efforts needed for mundane activities in SCM like contract management. Of course, there are course corrections to be done and blockchain is an evolving technology. But it gives a promising ray of hope in the world of Trade, Commerce, and Supply Chain Management.
References:
<ol>
<li><a href="https://hbr.org/2020/05/building-a-transparent-supply-chain">https://hbr.org/2020/05/building-a-transparent-supply-chain</a></li>
<li><a href="https://www.ibm.com/blockchain/supply-chain">https://www.ibm.com/blockchain/supply-chain</a></li>
<li><a href="https://www.mckinsey.com/business-functions/operations/our-insights/blockchain-technology-for-supply-chainsa-must-or-a-maybe">https://www.mckinsey.com/business-functions/operations/our-insights/blockchain-technology-for-supply-chainsa-must-or-a-maybe</a></li>
<li><a href="https://www2.deloitte.com/us/en/pages/operations/articles/blockchain-supply-chain-innovation.html">https://www2.deloitte.com/us/en/pages/operations/articles/blockchain-supply-chain-innovation.html</a></li>
</ol>
Thanks for reading!

From sourcing raw materials from every corner of the world and selling end products across the world, Supply chain management plays a vital role! And technology never failed to do its job! Blockchain has already disrupted the finance industry and the supply chain will be the next

Building the next-gen Supply Chain with Blockchain Technology

One of the challenges that we face while testing the application is testing it on different browsers and environments. We all might have observed that some websites are not properly displayed on some browsers and we just think that the website is broken. But, as soon as you open it on a different browser, the website opens up just fine. Thus, this behavior explains the compatibility of a website with different browsers.
<ul>
<li>Kiran HS, <a href="https://www.coffeebeans.io/">CoffeeBeans</a></li>
</ul>
<img src="https://media.coffeebeans.io/blogs/kiran-blog-tn.jpg" alt="cover">
Each browser interprets the information on the website page differently, so, some browsers may lack the features that your website is trying to show and make your website look broken on that browser.
That being the brief story, I bet you all have figured out the topic of today’s discussion — Cross Browser Testing. So what is cross browser testing?
Cross browser testing is a type of testing that lets you check whether the websites are working as intended when accessed through
<ul>
<li>Different Browser — OS combination</li>
<li>Browsers like — Chrome, Firefox, Safari, Edge, etc</li>
<li>Operating Systems like — Windows, Linux, MAC, etc</li>
</ul>
It’s about shipping releases that are as browser-agnostic as possible, which is key to delivering a uniform user experience on a diverse, ever-growing range of browsers/devices.
As a tester, it is your responsibility to make sure that not only do applications work, but they work for all your users, no matter what browser, device, or additional assistive tools they are using. Now comes the next question –
Why cross-browser testing?
<ol>
<li>To check how the website appears on different browsers</li>
<li>To check how the website works on different browsers ( quite obvious )</li>
<li>To ensure that you’re not alienating a significant part of your target audience–simply because your website does not work on their browser OS.</li>
</ol>
The next question which comes to mind is –
How to perform cross-browser testing?
This can be done in two (obvious) ways
<ol>
<li>Manual Approach (Manually check the functionality on different browsers and platforms, painful process to follow)</li>
<li>Automation Approach (Write one piece of code and run it on different browsers and platform combinations — sounds cool and easy right?)</li>
</ol>
That being said, let’s dive into the automation approach and see how this can be achieved with Selenium, Selenium GRID, and TestNG
Firstly, you will need to set up a Selenium grid to run the tests on different browsers and platforms. For this, you will need a Selenium server standalone jar which can be found in the given link — <a href="https://selenium-release.storage.googleapis.com/index.html?path=3.141%2F">Selenium Server Standalone Jar</a>
Once you have downloaded the jar file, you need to set up the hub and node based on the requirement. For the demonstration purpose, I will create 1 hub and 2 nodes. 1 of the nodes will run firefox as browser-related test cases. The other node will run chrome-related test cases.
Let’s start by creating a hub.
The command to create a node is as follows
java -jar selenium-server-standalone-3.141.59.jar -role hub
Below is the screenshot of the hub looks when it is up and running
<img src="https://media.coffeebeans.io/blogs/0_G_mWZ6m7Q0JdPGC4.png" alt="terminal"> 
Each browser interprets the information on the website page differently, so, some browsers may lack the features that your website is trying to show and make your website look broken on that browser.
That being the brief story, I bet you all have figured out the topic of today’s discussion — Cross Browser Testing. So what is cross browser testing?
Cross browser testing is a type of testing that lets you check whether the websites are working as intended when accessed through
<ul>
<li>Different Browser — OS combination</li>
<li>Browsers like — Chrome, Firefox, Safari, Edge, etc</li>
<li>Operating Systems like — Windows, Linux, MAC, etc</li>
</ul>
It’s about shipping releases that are as browser-agnostic as possible, which is key to delivering a uniform user experience on a diverse, ever-growing range of browsers/devices.
As a tester, it is your responsibility to make sure that not only do applications work, but they work for all your users, no matter what browser, device, or additional assistive tools they are using. Now comes the next question –
Why cross-browser testing?
<ol>
<li>To check how the website appears on different browsers</li>
<li>To check how the website works on different browsers ( quite obvious )</li>
<li>To ensure that you’re not alienating a significant part of your target audience–simply because your website does not work on their browser OS.</li>
</ol>
The next question which comes to mind is –
How to perform cross-browser testing?
This can be done in two (obvious) ways
<ol>
<li>Manual Approach (Manually check the functionality on different browsers and platforms, painful process to follow)</li>
<li>Automation Approach (Write one piece of code and run it on different browsers and platform combinations — sounds cool and easy right?)</li>
</ol>
That being said, let’s dive into the automation approach and see how this can be achieved with Selenium, Selenium GRID, and TestNG
Firstly, you will need to set up a Selenium grid to run the tests on different browsers and platforms. For this, you will need a Selenium server standalone jar which can be found in the given link — <a href="https://selenium-release.storage.googleapis.com/index.html?path=3.141%2F">Selenium Server Standalone Jar</a>
Once you have downloaded the jar file, you need to set up the hub and node based on the requirement. For the demonstration purpose, I will create 1 hub and 2 nodes. 1 of the nodes will run firefox as browser-related test cases. The other node will run chrome-related test cases.
Let’s start by creating a hub.
The command to create a node is as follows
java -jar selenium-server-standalone-3.141.59.jar -role hub
Below is the screenshot of the hub looks when it is up and running
<img src="https://media.coffeebeans.io/blogs/0_Q0WAC0s4LgUU164D.png" alt="terminal">
Once you have registered that node, let’s register another node with firefox browser configurations. The command to do so is as follows
java -Dwebdriver.gecko.driver=”/Users/macbook/Downloads/geckodriver” -jar selenium-server-standalone-3.141.59.jar -role node -hub <a href="http://localhost:4444/grid/register">http://localhost:4444/grid/register</a>
Below is the screenshot of it looks when it is registered
<img src="https://media.coffeebeans.io/blogs/0_CWb2UdLTXEpKqkQc.png" alt="terminal">
Once both the nodes are registered it can be verified by navigating to the grid console. The same can also be verified on the hub console
Hub Console
<img src="https://media.coffeebeans.io/blogs/0_oA_nApXOVaF-rIA1.png" alt="terminal">
Grid console URL is as follows — <a href="http://localhost:4444/grid/console">GRID Console</a>
<img src="https://media.coffeebeans.io/blogs/0_fOZGNwa3mrtlm72W.png" alt="GRID console">
After you have registered both nodes, get the hub URL and append wd/hub at the end. We will use this URL to route the test cases to specific nodes as configured.
Hub Url — <a href="http://localhost:4444/wd/hub/">Hub URL</a>
Now let’s create a maven project and testng.xml file so that we can run the tests parallelly.
Let’s add the selenium and testng dependency
<img src="https://media.coffeebeans.io/blogs/0_IeJ1q2hDcy8dRTh_.png" alt="image">
<img src="https://media.coffeebeans.io/blogs/0_6uKiJSq_bKiP8hyG.png" alt="image">
After you have registered both nodes, get the hub URL and append wd/hub at the end. We will use this URL to route the test cases to specific nodes as configured.
Hub Url — <a href="http://localhost:4444/wd/hub/">Hub URL</a>
Now let’s create a maven project and testng.xml file so that we can run the tests parallelly.
Let’s add the selenium and testng dependency
<img src="https://media.coffeebeans.io/blogs/0_jsHJ65hHx_7ZSKWT.png" alt="code editor">
<img src="https://media.coffeebeans.io/blogs/0_bBBUErzRNXYoTJ9Y.png" alt="code editor">
Now let’s run the test cases parallely and see if the configurations are working
Command to run the test — mvn clean test
Viola, one single test can now be run on 2 different browsers parallelly. Refer to the console for execution results
<img src="https://media.coffeebeans.io/blogs/0_xRAeer4-SBYGvmAD.png" alt="code editor">
For now, the platform in which I am running the test cases is MAC, this can be changed when a different platform is configured like Windows, Linux, etc. TestNG file to run the test cases parallely is as below.
<img src="https://media.coffeebeans.io/blogs/0_ZRKWiFBK8S87v787.png" alt="code editor">
In the same way, safari and edge can also be configured to run the test cases. It depends on the requirement of how many browsers and platforms need to be covered.
That’s all folks, for now, see you guys in the next exciting blog.
Thank you !!

One of the challenges that we face while testing the application is testing it on different browsers and environments. We all might have observed that some websites are not properly displayed on some browsers and we just think that the website is broken. But, as soon as you open it on a different browser, the website opens up just fine. Thus, this behavior explains the compatibility of a website with different browsers.

Cross Browser Testing using Selenium GRID

In today’s rapidly evolving threat landscape, organizations need robust security measures to protect their systems and data. Threat modeling is an essential process that helps identify and address potential vulnerabilities early in the software development life cycle. By combining the Security Burrito approach and the STRIDE model, organizations can enhance their threat modeling practices and strengthen their overall security posture. In this blog, we will explore how these two approaches work together to mitigate threats effectively.
<img src="https://media.coffeebeans.io/blogs/Enhancing-Security-1-1-1-2048x1152.jpg" alt="cover">Threat Modeling with the Security Burrito Approach
The Security Burrito approach, emphasizing continuous security throughout the project life cycle, provides a solid foundation for threat modeling. Here’s how it can be applied:
Incorporate Security from the Start: By integrating security considerations from the initial stages of design and development, organizations ensure that security is a fundamental aspect of the project. This includes identifying potential threats and considering countermeasures early on.
Continuous Security Assessment: Adopting a continuous security mindset allows for ongoing assessment of potential threats. Regular security reviews and risk assessments help identify and address vulnerabilities in real-time, reducing the likelihood of security breaches.
Collaboration and Communication: Encouraging collaboration between developers, security professionals, and other stakeholders fosters a proactive security culture. Effective communication ensures that threat modeling activities align with project goals and that potential threats are properly understood and addressed.
Applying the STRIDE Model in Threat Modeling
The STRIDE model is a valuable framework for identifying potential threats in software systems. It stands for the following threat categories:
Spoofing Identity: This includes threats such as impersonation or unauthorized access. Countermeasures may involve implementing strong authentication mechanisms, multi-factor authentication, and robust user identity management.
Tampering with Data: Threats in this category involve unauthorized modification or manipulation of data. Countermeasures can include data validation, input sanitization, and encryption to protect data integrity.
Repudiation: This category focuses on threats related to denying or disputing actions or events. Implementing audit logs, digital signatures, and secure timestamps helps establish non-repudiation and traceability.
Information Disclosure: Threats in this category pertain to unauthorized access or exposure of sensitive information. Countermeasures may involve data encryption, access controls, and secure transmission protocols.
Denial of Service: These threats aim to disrupt or disable system functionality. Countermeasures may include implementing rate limiting, traffic monitoring, and employing mitigation strategies against DoS attacks.
Elevation of Privilege: This category deals with unauthorized access to elevated privileges. Countermeasures may involve implementing strong access controls, privilege separation, and least privilege principles.
Combining the Security Burrito approach with the STRIDE model enhances threat modeling practices:
Continuous threat identification and mitigation: By continuously assessing threats and vulnerabilities, organizations can promptly identify and address security issues using the appropriate STRIDE categories.
Proactive security measures: By integrating security from the early stages and fostering a security-first mindset, organizations can proactively implement countermeasures to mitigate identified threats effectively.
Collaboration and knowledge sharing: The Security Burrito approach promotes collaboration between stakeholders, allowing for collective understanding and action against identified threats based on the STRIDE model.
Conclusion
Threat modeling is a critical component of effective security practices. By combining the Security Burrito approach with the STRIDE model, organizations can strengthen their threat modeling efforts and enhance their overall security posture. This integrated approach ensures that potential threats are identified and mitigated throughout the project life cycle, enabling organizations to build more secure and resilient software systems.
Credits: Lok Bahadur, Software Engineer.

In today’s rapidly evolving threat landscape, organizations need robust security measures to protect their systems and data. Threat modeling is an essential process that helps identify and address potential vulnerabilities early in the software development life cycle. By combining the Security Burrito approach and the STRIDE model, organizations can enhance their threat modeling practices and strengthen their overall security posture. In this blog, we will explore how these two approaches work together to mitigate threats effectively.

Enhancing Security with Threat Modeling Using the Security Burrito Approach and STRIDE

1) In today’s technology landscape, companies are facing increasing pressure to deliver software quickly and efficiently. However, the challenge of managing complex infrastructure can create friction between development and operations teams, leading to inefficiencies and delays in the development process.
Please tell us the challenges faced due to this friction from your experience.
Interesting to start with this part. This rift between the development and operations team is actually what is called the DevOps divide. 
The most obvious one is a delay in the delivery. With a larger DevOps divide, communication and collaboration is sub-optimal. This can be further worsened if both teams have a difference in vision/priorities/strategies leading to further misunderstanding. With so much difference between the two core teams, the delivery will be riddled with bugs, issues, gaps, process leaks, and delays. 
One other challenge I have faced very often as a DevOps manager is a lack of visibility. Outside the team, not many people are even aware of what is happening. A true form of a closed black box. it is so close you have access to it but it is so opaque that you think nothing happens. This also causes issues with people not that familiar with the operations part as they think no work is happening. When one does not understand infrastructure operations that well, we tend to map it to regular applications. but status updates/jira tickets are and cannot be on the same level. And suddenly no one understands what is happening which over time results in, I don’t know what the Ops team is doing. And now you are left to fight the perception that you are working on something. 
However, the one challenge that is a force multiplier is that this DevOps divide results in finger-pointing and blaming when things don’t work as is. The devs team often says the environment is always unstable hence the productivity was impacted. The ops team says that the logs were clear, you should have just gone and fixed it. The Dev team thinks it is not my work as it is the work of a DevOps engineer and this goes on and on. This finger-pointing goes beyond one incident and often keeps the team in the ‘Storming’ phase in the Tuckerman model. When it hits this you are no longer only dealing with unstable environments but also with people who are not too keen to work together. 
2) What according to you are some of the most common sources of friction between development teams and infrastructure teams?
Incident response and RCA’s are probably the most common source of friction IMO. Actually, it is not directly that. Incident response is a symptom of a much deeper problem i.e. a sense of ownership of the stability of the environment. While the ops team works on creating the infra and the dev teams build the application, these pipelines are what keep the system running. One fault on it means a direct impact and the productivity dips. As serious as it sounds, I have often seen more of where the DevOps person is to fix this and less of, hey let me get my hands dirty. A counterpart to it could be the lack of skill but c’mon, at one point in time all of us were proficient in only one stack. We then evolved. But DevOps comes across as a scary world to many of the devs I have worked with who just literally and figuratively freeze when the pipelines fail. They don’t want to go to the deeper darker world of DevOps 😀 
The other side of this coin is that in many teams the Dev teams are not enabled with enough information to be able to fix the issues. Sometimes it is just expertise and sometimes there is a lack of a how-to. 
We need our DevOps to build environments in such a way that we know, acknowledge, and work towards the idea that one day this environment will be maintained by a developer who doesn’t have god/super admin access.
3) Can you share any examples of how friction between these teams has negatively impacted a project or a company?
What DevOps does is usually the one behind the scenes. They build the backbone of the entire application. You wouldn’t see a state-of-the-art UI as an output to show. but it will ensure that the API’s are not hacked. You will not see database entries in millions of records but it will ensure that all resources scale enough at the right time for those millions of records to be inserted. DevOps usually comes into focus when things break. 
There are plenty of examples of how the divide can lead to a bad reputation for an organization. 
Longer downtime, no disaster recovery planned, Security vulnerabilities, Truck factors of 1 or 2, Innovation dies down as we are always firefighting. 
Let’s say that the dev team plans a rollout with some eye-catching code without informing the Ops team. This leads to increased sales which is great for the company but now with the Ops team in the blind, the infra has not been revamped to handle the new scale. HPA and ASG are not just click of buttons. As a result of this, the site crashes leading to bad business values. 
Another classic example is cloud migration. Anytime we speak of this, the development teams say that this is not on us the DevOps team will take care. But in reality, it is a combined effort as the DevOps team needs to know what to migrate. Without this knowledge from the development team, it is possible that some things are left out, and hence post-migration, there is a downtime, bad user experience, or just significant delays in the migration as it is not working. 
Another thing is a hotfix/patch deployment. When a critical bug is reported, the development team rushes to fix it as it causes an impact on the revenue. While it is busy on this, the DevOps team is unaware and is sipping pina coladas on the beach. Suddenly when all is fixed, the DevOps team gets a message to deploy ‘right away’ as it is a patch. This is immense pressure on the DevOps team who until a few seconds ago were unaware. While emergency patches are unavoidable, including the DevOps team from the beginning helps in reducing the pressure and planning much better. 
4) When it comes to reducing friction, what are some strategies that have worked well in your experience?
The one thing that worked wonderfully well was building a cross-functional team. At CoffeeBeans, we invested time into making DevOps a mindset across the org. All devs had to undergo mandatory training to learn about the basics of ci/cd pipelines. This gave them a vast understanding of how things were built and removed the fear of the big black box. We also had DevOps 101 for the non-technical teams such as business analysts, and talent acquisition so that they know what is expected when we say DevOps and everything related. 
The other tiny effort that made a lot of difference is having a slightly different format of updates in Jira cards. We did not talk about Given-When-Then, we spoke about which environment, what is the downtime, and who are the stakeholders to be informed. We retained Acceptance criteria though. 
Another thing that worked and helped immensely when new people were on board was having a standard set of questions in a checklist for every application we built. This ensured that we were maintaining standardization and consistency across projects and most importantly, we were not missing out on anything critical. This list also helped us chalk out very clearly what the project needs and what it does not. a crucial point that helped us with our client communications. 
5) We often hear about the importance of balancing agility and innovation with stability and reliability. Can you share any insights on how to do this effectively in a complex infrastructure environment?
<ol>
<li>The mantra of any application now is CI/CD. Once setup, this is a significant game changer in the way agility can be infused with stability </li>
<li>Define the vision/priorities/goals and ensure both teams understand that. </li>
<li>Draw a virtual line between components in an infrastructure based on criticality. Must-go ones have to be in a stable state all the time. the other segments that can afford a shift/downtime can be used in innovation and playground</li>
<li>Use IaC</li>
<li>different deployment strategies like A/B, the canary that let you reduce the risk of downtime giving you enough space to explore and innovate</li>
<li>Exhaustive monitoring and alerting</li>
</ol>
6) What are some potential risks or pitfalls to avoid when trying to reduce friction between development teams and infrastructure teams?
<ol>
<li>buy-in from the leadership – Make sure your leadership knows the why behind the what of changes in the DevOps strategies. </li>
<li>Assuming everyone knows – don’t assume everyone knows the reason/rationale/approach behind changes or even that it is happening. Communicate</li>
<li>Not addressing tech debts soon enough</li>
<li>Wanting everything yesterday – This is a double-edged sword. With the time to market being the least ever, it is common for everyone to expect all deliverables yesterday. and for the things we don’t understand, we often tend to underestimate the complexity and time needed. We should leave the work to the experts and have due faith in that. </li>
<li>Unclear processes for change management </li>
<li>Undefined SLAs in change management</li>
</ol>
7) Finally, can you offer any advice or best practices for organizations looking to improve collaboration and reduce friction between these teams?
<ol>
<li>Invest in automation. </li>
<li>Give space and time to clear out the tech debts. This goes a long way in ensuring the rigidity and reliability of the system. </li>
<li>Acknowledge the cultural and mindset differences between the development and operations teams. Their work expects different kinds of working styles. Even their format of Jira updates wouldn’t be the same. </li>
<li>don’t downplay it when someone says it takes me that long to build an environment</li>
<li>if the pipelines are red for long and red often, the root cause is something other than a devops people being a truck factor. </li>
</ol>
<h2>Conclusion</h2>
Reducing friction between development teams and complex infrastructure requires a combination of technical solutions and cultural changes. By adopting DevOps practices, automating infrastructure, using containerization, fostering communication and collaboration, and providing adequate training, you can ensure that both teams are working together towards the same goal, which is to deliver high-quality software quickly and efficiently. With the right approach, organizations can reduce friction, improve efficiency, and deliver software faster and more effectively.

1) In today’s technology landscape, companies are facing increasing pressure to deliver software quickly and efficiently. However, the challenge of managing complex infrastructure can create friction between development and operations teams, leading to inefficiencies and delays in the development process.

Reducing Friction between Development Teams and Complex Infrastructure: Interview with Dhivya, Head of Delivery

If you are a performance engineer/network engineer or even security engineer, the chance of you encountering eBPF technology in the future is very high. eBPF now has a huge community of users, including big players like Meta, Google, Cloudflare, and Netflix all using this tech in their daily operations.<img src="https://media.coffeebeans.io/blogs/1_PAHzzTuVSwAHRXwIzOFjOw.jpg" alt="cover">
<h2>Prelude</h2>
Let me start the blog with a real story. A year back one of my friends called me to discuss tech (which is a very common thing between us). We share different technical challenges each of us faces at our workplace or by any of our peers and these discussions lead to some informative and creative knowledge-sharing sessions. In such a discussion, he described a specific challenge faced by his cousin, who works for a giant cloud provider. The challenge was to restrict certain IPs dynamically as they provide threats such as a DOS (Denial of service attack), the application developer’s brain in me impetuously replied that these should be handled at a firewall level, or middleware can be written to check the origin of the packet and maintain a blacklist for the malicious senders and ignore the requests (Yes I come from a NodeJS and Go background so the initial solution strikes as a middleware). My friend patiently explained the scale and performance at which this needed to be executed which was way beyond my comprehension. After a noob’s doubt clearing session, we agreed that the scale he wanted could only be achieved at a kernel level. I wished him luck (sarcastically) to write a kernel patch and raise a PR hoping the OS maintainers would include the kernel patch in an upcoming kernel release and he can use this feature when it is released. As a reply to my sarcasm, he shared with me a link to an article that detailed something called “eBPF” (extended Berkeley Packet Filter). I did a basic skimming through the article, and my ignorant mind came to sense that there are amazing inventions in the tech world that I am unaware of.
According to eBPF, you can inject the code directly into the kernel without writing a patch, waiting for it to be approved by the OS maintainer
“RUN YOUR CUSTOM CODE DIRECTLY IN OS KERNEL” — LIZ RICE
“SUPERPOWERS FOR LINUX”. — BRENDAN GREGGS<img src="https://media.coffeebeans.io/0_17I5uYtBRc_AXsy8.png" alt="image">NB: I have added some video and blog links in the reference section please check it out for some amazing sessions and blogs of eBPF.
<h1>History</h1>
The eBPF came to life in 2014, introduced in Linux kernel 3.18, thereby unlocking the God mode of the Linux kernel. The natural doubt anyone reading this blog would have is regarding the name. If this is an “extended berkeley packet filter” then there should be a BPF “berkeley packet filter”. Well, you are right. The BSD packet filter is not a new concept. It was from the 90’s. This gem has been hiding under the radar for years, the Xennails were true innovators. BPF was very basic and its only job was to filter packets at the kernel level hence the name.
NB: I have added the original BPF paper published on December 19, 1992, a very interesting read.
The eBPF has come a long way from BPF, just a packet filtering utility to the consideration of microservices architecture for kernels or as they call it microkernels. All the top tech companies that work at scale nowadays use eBPF on daily basis. CNCF community nowadays breathes and lives on eBPF, if you are a DevOps engineer or sysadmin you would have heard of cilium and Falco both popular in Kubernetes users and production tools that are written on top of eBPF. In 2018 Linux announced it would replace its iptable-based implementation with an eBPF version in the kernel (well replacing iptable with any solution would be better), fall back and disadvantage of using iptables is out of the scope of this article, please go to the reference section and find a well-written article about it. The Kubernetes used iptables for the following use cases mostly
<ul>
<li>Kube-proxy — the component which implements Services and load balancing by DNAT iptables rules</li>
<li>Most CNI plugins are using iptables for Network Policies</li>
</ul>
Cilium has made it more efficient by eliminating the iptable whose performance degrades. 
<ol>
<li></li>
</ol>
Program Execution Bozo’s Guide
To explain the importance of eBPF there needs to be an explanation of how programs are executed in Linux, I will try to explain it from a 1000ft view for everyone.
NB: Windows User? Well why are you even reading this article, you guys do not have all these cool features.
Linux memory is divided into two
<ol>
<li>Kernel space</li>
<li>User space<img src="https://media.coffeebeans.io/blogs/0_ErgxMOK9ov19toqQ.jpg" alt="image">The image itself explains the difference between these two. All the programs that you write are just collections of syscalls that are kernel APIs. Just take the example of opening a file through your favorite programming language that just translates into a fileopen syscalls in the kernel.</li>
<li>When your application asks the kernel for something, a chunk of data in kernel space is frequently copied into user space. We must do this because operating systems strictly partition memory regions used by the kernel, making it impossible to simply provide a pointer to some region of kernel memory to a user space program. This is known as “crossing the user/kernel boundary,” Because of the copy operation, operations like these can have significant performance consequences.</li>
<li>While syscalls cover almost all cases, there arise situations where this is not sufficient like when we need kernel-level performance or write a new driver programming, etc. Depending on the OS maintainers to make patches for all these small use cases is a waste of time and an impossible process. This is where ebpf comes into the picture.</li>
<li>eBPF helps you to write programs in the user space which get packaged and injected into the kernel directly, these programs run on VM in the kernel with a limited instruction set thus extending the capability of the base kernel module.</li>
</ol>
<h1>eBPF Dissected</h1>
eBPF is the provision to run custom code that runs on the kernel for various processes like
<ul>
<li>Observability (tracing)</li>
<li>Debugging</li>
<li>Firewalling</li>
<li>Load Balancing</li>
<li>Network related activity</li>
</ul>
Anyone who has worked in tracing the various programs in the kernel would know its difficulty. The half-baked utilities available in the Linux systems are not enough for profile complex systems or even to extend the perf tooling.
Ebpf is event-driven which means it gets triggered on the following scenarios
<ul>
<li>A system call</li>
<li>Function entry/exit</li>
<li>When a packet enters or leaves</li>
<li>K probes or U probes</li>
</ul>
The programs are written in a language called restricted c which is c with a limited instruction set. The BPF compiler BCC converts this into a bytecode which is loaded into the kernel for execution. A validator is run before compiling to ensure there is no infinite loop or such never-ending I/O operation which could crash the kernel.
<h1>Additional Trick Under Your Sleeve</h1>
The ebpf is indeed a powerful tool that you could have under your sleeve. When working on high-performance projects tweaking the packets or extending the tracing functionality all help you give better observability of what’s happening with the system. Even though encountering the ebpf by an application developer at the present stage is very feeble, if you are a performance engineer/network engineer or even security engineer, the chance of you encountering ebpf in the future is going to shoot up to the sky.
There are some considerations while writing ebpf programs, there have been several privilege escalation attacks that leverage ebpf since it runs in a sudo privilege. The ebpf programs could be used as a powerful aid when leveraging kernel memory vulnerabilities. A detailed writeup of leveraging such a vulnerability was found by Qualys, there is a writeup by them which you can refer to from <a href="https://www.qualys.com/2021/07/20/cve-2021-33909/sequoia-local-privilege-escalation-linux.txt">here</a>.
<h1>Conclusion</h1>
As said in Spiderman movies “Great power comes with great responsibility” when you unlock the God mode of Linux you are on your own, the guards that protected your program from corrupting the whole are not available now. There are specific use cases to use Ebpf, it is not the swiss knife for all your performance issues. The community is pretty huge now including big players like meta, google, Cloudflare, and Netflix all using the tech daily. The tech has loads of potential to grow, recent years have seen separate conf for ebpf enthusiasts.
This blog serves as a small opening to people who are unaware of this cool tech, so please do your research. There are tons of resources available online about ebpf and open source projects being built on top of it. I will be writing a follow-up article detailing how to write a sample ebpf program and execute it.

If you are a performance engineer/network engineer or even security engineer, the chance of you encountering eBPF technology in the future is very high. eBPF now has a huge community of users, including big players like Meta, Google, Cloudflare, and Netflix all using this tech in their daily operations.

eBPF — Divulging The Hidden Super Power

<h1>Appium setup made easy — Step by step process for Installing/ Configuring Appium &#x26; Appium Doctor on macOS for Android platform</h1>
— Vikas Kumar Garg, <a href="https://coffeebeans.io/">CoffeeBeans Consulting</a>
<img src="https://media.coffeebeans.io/blogs/1_c3OH6POKO6PAHEpn0kKENQ.jpg" alt="cover">
<h1>What Is Appium</h1>
Appium is an open-source tool for automating native, mobile web, and hybrid applications on iOS mobile, Android mobile, and Windows Native apps that are written using the iOS, Android, or Windows SDKs.
Mobile web apps are web apps accessed using a mobile browser (Appium supports Safari on iOS and Chrome or the built-in ‘Browser’ app on Android). Hybrid apps have a wrapper around a “web-view” — a native control that enables interaction with web content.
Appium is a “cross-platform” tool: it allows you to write tests against multiple platforms (iOS, Android, Windows), using the same API. This enables code reuse between iOS, Android, and Windows test suites.
<img src="https://media.coffeebeans.io/blogs/0_sKN3XXpV81_7KuD0.png" alt="appium setup">
Appium + Appium Doctor
This article explains in detail how to install &#x26; Configure Appium on MAC with Android studio, related tools, and dependencies for Android.
This is a complete guide so you do not need to do anything separately for configuring Appium for Android on macOS. By using Appium-doctor we will check if our setup is done right. Different sections have been added to the article to explain different components.
<h1>Step-1: Install JAVA JDK</h1>
Java is required to run Appium desktop as well as Appium command line.
You should go to the below link and install JAVA JDK for MAC OS.
<img src="https://media.coffeebeans.io/blogs/1_KmGBWfZ4xDTkdc5ivwLR4g.png" alt="Java SE 11">
<a href="https://www.oracle.com/in/java/technologies/javase/jdk11-archive-downloads.html">https://www.oracle.com/in/java/technologies/javase/jdk11-archive-downloads.html</a>
<h1>Step-2: Set JAVA_HOME</h1>
Appium setup require JAVA_HOME to be set so that it can detect java related requirements while running Appium-Server/ Appium-Doctor.
Open a new terminal window and open .zprofile file by typing: vim .zprofile/ vi .zprofile now type below lines in the file:
export JAVA_HOME=/Library/Java/JavaVirtualMachines/jdk-11.0.13.jdk/Contents/Home
Then, press esc key and enter :wq in the .zprofile to save the file
Open a new terminal and run the below command:
>JAVA_HOME/bin/java -version
You should see this –
<img src="https://media.coffeebeans.io/blogs/0_U8gxJv9y5Ze8Hir5.png" alt="terminal">
Now check JAVA_HOME by typing >echo $JAVA_HOME. You should be able to see below mentioned screenshot :
<img src="https://media.coffeebeans.io/blogs/0_pTxHuk-wU4zf81BS.png" alt="terminal">
<h1>Step-3: Install Android Studio / SDK tools:</h1>
For running Appium tests we need android SDK, Emulators and other related libraries. Android studio provides all these required libraries and emulators as well.
Go to <a href="https://developer.android.com/studio/">https://developer.android.com/studio/</a> and download latest version of android studio.
Launch the .DMG file as installer and the Android Studio Setup Wizard guides you through the rest of the setup, which includes downloading Android SDK components that are required for the setup.
After Opening and selecting the default settings, Android Studio will start downloading the required libraries as shown below.
<img src="https://media.coffeebeans.io/blogs/0_QhjUSgCXjtPkBotM.png" alt="android studio">
<h1>Step-4: Set ANDROID_HOME:</h1>
We should setup all below mentioned environment variables as they are needed by Appium server as well as by Appium-doctor while running the tests and inbuilt applications.
export ANDROID_HOME=/Users/&#x3C;username>/Library/Android/sdk
export PATH=$PATH:$ANDROID_HOME/platform-tools
export PATH=$PATH:$ANDROID_HOME/platform-tools
export PATH=$PATH:$ANDROID_HOME/tools/bin
export PATH=$PATH:$ANDROID_HOME/tools/bin
It looks like this in my machine as shown below in the screenshot
<img src="https://media.coffeebeans.io/blogs/0_MYkc4Kl23Awwybmg.png" alt="terminal">
Now Type this command in the terminal : >android — You should see as shown in this screenshot:
<img src="https://media.coffeebeans.io/blogs/0_iM8xrf7bYwcV-_cd.png" alt="terminal">
You can check the ANDROID_HOME path by typing this command:
>echo $ANDROID_HOME
<img src="https://media.coffeebeans.io/blogs/0_RLh9yqjBIXe783yv.png" alt="terminal">
<h1>Step-5: Install NodeJS and NPM:</h1>
For installing Appium and other Appium related application via terminal we need nodejs and npm.
Check if nodejs is already installed : >node -v
Check if npm is already installed : >npm -v.
If it is installed you will see as shown below in the screenshot
<img src="https://media.coffeebeans.io/blogs/0_hROT5-a4Z-Uaxk6M.png" alt="treminal">
If node and NPM are not installed then go to — <a href="https://nodejs.org/en/download/">https://nodejs.org/en/download/</a> &#x26; download mac installer as shown below.
<img src="https://media.coffeebeans.io/blogs/0_-81SGGumBIuLEuQZ.png" alt="Appium setup GUI">
<img src="https://media.coffeebeans.io/0_mxeAM-Kv51BBJSjx.png" alt="android studio">
<h1>Step-6: Install Appium using NPM via terminal</h1>
<pre><code class="language-shell">> npm install -g appium #installs appium> npm install wd # installs appium web clientIf there is permission issue - use sudo before above commands as shown in the screenshot below.
</code></pre>
<h1>Step-7: Install Appium Desktop</h1>
Appium Desktop is used to configure Appium inspector and other dependencies which are not directly supported in Appium installed via npm. It provides a UI where we can set the dependencies.
Go to this link: <a href="https://github.com/appium/appium-desktop/releases/tag/v1.22.2">https://github.com/appium/appium-desktop/releases/</a> and download dmg installer for mac.
<img src="https://media.coffeebeans.io/blogs/0_GZTjYspmDP5arT2Y.png" alt="terminal">
Click and download .DMG file. For example, “ <a href="https://github.com/appium/appium-desktop/releases/download/v1.22.2/Appium-Server-GUI-mac-1.22.2.dmg">Appium-Server-GUI-mac-1.22.2.dmg</a>“. Double click the .DMG file When the download finishes, and drag and drop the Appium into Applications as shown below.
<img src="https://media.coffeebeans.io/blogs/0_WdxR9RUDNYhnHFzl.png" alt="github">
You should be able to search Appium via spotlight search[command+shift key] in MAC OS as shown below:
<img src="https://media.coffeebeans.io/blogs/0_a89FifbAwSFtqAmU.png" alt="Appium server GUI">
Click on Appium Server GUI → Some people might see the popup as shown below:
<img src="https://media.coffeebeans.io/blogs/0_GFmzgvQV2cPkU88P.png" alt="appiu">
Click ok on the screen and then open system preferences as shown below
<img src="https://media.coffeebeans.io/blogs/0_BxTR1wYMbtJ_Uvnh.png" alt="mac os alert popup">
Click on “Security and Privacy” as highlighted in above screenshot.
You should see as shown below in the screenshot.
<img src="https://media.coffeebeans.io/blogs/0_3IwvfEVEb-MLHIXX.png" alt="settings">
Click on “Open Anyway” in above screenshot and you should see the screen as shown below:
<img src="https://media.coffeebeans.io/blogs/0_U92kfxkGBQT2swnU.png" alt="system preference">
<img src="https://media.coffeebeans.io/blogs/0_kvhlDIcTciYUc-43.png" alt="system preference">
Click on the button “Open” in above screenshot and Appium desktop server startup screen will be shown as displayed below:
<img src="https://media.coffeebeans.io/blogs/0_CnuZH5nI4_EhFWPs.png" alt="mac os popup">
Click on “Start Server v1.22.2” in above screenshot and appium server will start and following screen will appear:
<img src="https://media.coffeebeans.io/blogs/0_MQ0OGM77Ao5TPuJC.png" alt="appium app">
Server will be stopped:
<img src="https://media.coffeebeans.io/blogs/0_AdxSotUobWbIaQhb.png" alt="appium server">
Click on Close Log button in the below screenshot and appium startup page will appear again.
<img src="https://media.coffeebeans.io/blogs/0_KhCG_QqXzG4BgpfJ.png" alt="appium server">
<img src="https://media.coffeebeans.io/blogs/0_S47H2U2KWZhbAACR.png" alt="appium app">
<h1>Step-8: Install Appium-Doctor</h1>
Appium doctor is a command line utility which helps in identifying setup related issues. It helps us in identifying whether our appium setup is proper or not. Appium doctor can be install via terminal using following command(use sudo if you face permission issue)
<img src="https://media.coffeebeans.io/blogs/0_qWfMTtdDg7t9K3MZ.png" alt="appium docker">
> npm install -g appium-doctor
<img src="https://media.coffeebeans.io/blogs/0_eieWK4hYcrRRH-pW.png" alt="terminal">
<h1>Step-9: Run Appium Doctor</h1>
Appium doctor helps us to make sure that our setup is working fine.
Run Appium doctor by following command: > appium-doctor android (For android setup verification). You will see the screen as shown below.
<img src="https://media.coffeebeans.io/blogs/0_yxHI18jcSnkHM5h2.png" alt="terminal">
The above screen confirms that Appium is set up properly for Android.
Stay Tuned for the more insightful blogs

Appium is an open-source tool for automating native, mobile web, and hybrid applications on iOS mobile, Android mobile, and Windows Native apps that are written using the iOS, Android, or Windows SDKs.  Mobile web apps are web apps accessed using a mobile browser (Appium supports Safari on iOS and Chrome or the built-in ‘Browser’ app on Android). Hybrid apps have a wrapper around a “web-view” — a native control that enables interaction with web content.

Installing/Configuring Appium on MAC OS for Android

<img src="https://media.coffeebeans.io/blogs/1_n6ZZJWEbyVCFfkU5ZjYMAA.jpg" alt="cover">
Activation functions are functions which are used in the Artificial Neural Networks to capture the complexities inside the data. A neural network without an activation function is just a simple regression model. The activation function does the non-linear transformation to the input making it capable to learn and perform more complex tasks.We introduce non-linearity in each layer through activation functions.
<img src="https://media.coffeebeans.io/blogs/0_Sn7tvyAqn0ZLxsMA.png" alt="neural network">
Let us assume there are 3 hidden layers, 1 input and 1 output layer.
W1-Weight matrix between Input layer and first hidden layer
W2-Weight matrix between first hidden layer and second hidden layer
W3-Weight matrix between second hidden layer and third hidden layer
W4-Weight matrix between third hidden layer and output layer
Below mentioned equations represents a feedforward neural network.
<img src="https://media.coffeebeans.io/blogs/0_o236kGYfgpPHW7mx.png" alt="formulas">
If we stack multiple layers, we can see output layer as a function:
<img src="https://media.coffeebeans.io/blogs/0_yMKRr-NxncVRCOM5.png" alt="formula">
What are Ideal qualities of an activation function:
The activation function generally introduce non-linearity in the network to capture the complex relations between input features and output variable/class.
2. Continuously differentiable:
The activation function needs to be differentiable since neural networks are generally trained using gradient descent process or to enable gradient based optimization methods.
3. Zero centered:
Zero centered activations functions makes sure that mean activation value is around 0. This is important because convergence is usually seen faster on normalized data. I have explained many of the commonly used activation below, some are zero centered some are not. Mostly when we have a activation function which is not zero centered we tend to use normalization layers like batch normalization to mitigate this issue.
4. Computational expense should be low:
Activation functions are used in each layer of the network and is computed a lot of times, hence its computation should be easy and not very computationally expensive.
5. Killing gradients:
Activation functions like sigmoid has a saturation problem where the value doesn’t change much for large negative and large positive values.
The derivative of the sigmoid function gets very small there which in turn prevents the updating of the weights in initial layers during backpropagation and hence the network doesn’t learn effectively. This should be avoided to learn patterns in the data and hence the activation function should not ideally suffer from this issue.
Most commonly used activation functions:
In this section we will go over different activation functions.
The sigmoid function is defined as:
The sigmoid function is a type of activation function which has a characteristic “S” shaped curve which has domain of all real numbers and output between 0 and 1. An undesirable property of the sigmoid function is that the activation of the neuron saturates either at 0 or 1 when the input from the neuron is either large positive and large negative. It is also non-zero centered which makes neural network learning difficult. In almost majority of the cases, it is always better to use Tanh activation function instead of sigmoid activation function.
2. Tanh function–
<img src="https://media.coffeebeans.io/blogs/0_Td9y1hYUN0TVKDRd.png" alt="tanh curve">
tanh curve
Tanh has just one advantage over sigmoid function that it is zero-centered and it’s value is binded between -1 and 1.
3. RELU(Rectified Linear Unit)-
<img src="https://media.coffeebeans.io/blogs/0_38oamhtIXD96wEvE.png" alt="RELU plot">
RELU plot
RELU is one of the many non zero-centered activation function and given this disadvantage it is still widely used because of the advantages it has. It is computationally very inexpensive, does not cause saturation and does not cause the vanishing gradient problem. The RELU function doesn’t have a higher limit, hence it has a problem of exploding activations and on the other hand for negative values, it has 0 activation and hence it completely ignores the nodes with negative values. Hence it suffers from “dying relu” problem.
Dying ReLU problem: During the backpropagation process, the weights and biases for some neurons are not updated because its nature where activation is zero for negative values. This might create dead neurons which never get activated.
4. Leaky RELU–
Leaky RELU is a type of activation function based on RELU function with a small slope for negative values instead of zero.
<img src="https://media.coffeebeans.io/blogs/0_NflMZDRuzHzGskcN.png" alt="Leaky RELU function">
Leaky RELU function
Here, alpha is generally set to 0.01. It solves the “dying RELU” problem and also its value is generally small and is not set near to 1 since it will only be a linear function then.
If we use alpha as hyperparameter for each neuron, it becomes a PReLU or parametrized RELU function.
5. ReLU6-
This version of ReLU function is basically a ReLU function restricted on the positive side.
<img src="https://media.coffeebeans.io/blogs/0_wngoFXXl5o8CcX6X.png" alt="ReLU6">
This helps in containing the activation function for large input positive values and hence stops the gradient to go to inf value.
6. Exponential Linear Units (ELUs) Function-
Exponential Linear Unit is also a version of ReLU that modifies the slope of the negative part of the function.
This activation function also avoids dead ReLU problem but it has exploding gradient problem because of no constraint on the activations for large positive values.
7. Softmax activation function-
<img src="https://media.coffeebeans.io/blogs/0_Utufr76XkwKTfUg1.png" alt="function">
It often used in the last activation layer of a neural network to normalize the output of a network to a probability value that in turn is mapped to each class which helps us in deciding the probability of output belonging to each class with respect to given inputs. It is popularly used for multi-class classification problems.
I hope you enjoyed reading this. I have tried to cover many of the activation functions which are commonly used in Neural Networks. If any mistake found, please feel free to mention and the blog will be corrected.

Activation functions are functions which are used in the Artificial Neural Networks to capture the complexities inside the data. A neural network without an activation function is just a simple regression model. The activation function does the non-linear transformation to the input making it capable to learn and perform more complex tasks.We introduce non-linearity in each layer through activation functions.

What are activation functions and why do we need those?

In today’s fast-paced digital world, providing efficient search functionality is crucial for delivering a seamless user experience. Algolia, a powerful search and discovery API, offers developers a way to implement robust search capabilities into their applications. In this blog post, we will explore how to integrate and synchronize Algolia with Strapi (v4) to enhance search functionality for a “Product” collection
<img src="https://media.coffeebeans.io/blogs/BLog-thumbnail-algolia.jpg" alt="cover">
Setting Up Algolia and Strapi Integration:
To begin, we need to set up Algolia and Strapi integration. We assume you already have a Strapi project and have installed Algolia. Now, let’s create a helper function that will initialize the Algolia index.
<pre><code class="language-javascript">const algoliasearch = require("algoliasearch");

const client = algoliasearch(
 process.env.ALGOLIA_APP_ID,
 process.env.ALGOLIA_API_KEY
);

const getAlgoliaIndex = (name) => {
 const index = client.initIndex(`${name}(${process.env.NODE_ENV})`);
 return index;
};

module.exports = getAlgoliaIndex;
</code></pre>
This code sets up the Algolia client using your Algolia app ID and API key. The getAlgoliaIndex function initializes the index with the provided name and environment variables. Note that we have also used the current environment with the index name to differ between the indices of different environments.
Synchronizing Strapi Product Collection with Algolia:
Now, let’s proceed to add and synchronize the “Product” collection in Strapi with Algolia. We will utilize Strapi’s lifecycle hooks to achieve this. Specifically, we will use the afterCreate lifecycle method.
Navigate to the product content type in your Strapi project and locate the lifecycles.js file. If it doesn’t exist, create one in the same directory. Add the following code to the lifecycles.js file:
<pre><code class="language-javascript">const getAlgoliaIndex = require("./path/to/getAlgoliaIndex");

const index = getAlgoliaIndex("product");

module.exports = {
 async afterCreate(entry, data) {
 const { id } = data.result;
 // Query product added to database
 const product = await strapi.db.query('api::product.product').findOne({
 where: {
 id
 }
 })

 try {
 // Save the object to Algolia
 await index.saveObject({ ObjectID: id, ...product });
 console.log("Object saved to Algolia");
 } catch (error) {
 console.error("Error saving object to Algolia", error);
 }
 },
};
</code></pre>
In this code snippet we import the getAlgoliaIndex function we created and then initialize the Algolia index by calling the function and passing a name for the index.
In the afterCreate lifecycle method we query the data which is added to the database.
Finally we use index.saveObject method to save the new entry to Algolia. We wrap this operation in a try-catch block to handle any potential errors.
Conclusion:
By integrating and synchronizing Algolia with Strapi, we have empowered our application with advanced search capabilities. Leveraging Algolia’s powerful search API, we can deliver fast and relevant search results to our users. This integration enables real-time updates to the Algolia index whenever a new product is created in Strapi. Implementing Algolia within Strapi not only enhances search functionality but also provides a scalable solution for managing search operations in your application.

In today’s fast-paced digital world, providing efficient search functionality is crucial for delivering a seamless user experience. Algolia, a powerful search and discovery API, offers developers a way to implement robust search capabilities into their applications. In this blog post, we will explore how to integrate and synchronize Algolia with Strapi (v4) to enhance search functionality for a “Product” collection

Integrating and Synchronizing Algolia with Strapi (v4)

Transfer Learning works on the idea that a pre-trained model trained in one setting can be utilized in a different setting for a specific task. I will be doing sentiment analysis here by using a pre-trained BERT model using Pre-Trained Model Approach and will be further training it on a custom dataset prepared.
<ul>
<li>By <a href="https://www.linkedin.com/in/lawa-747151117/">Lakshay Wadhwa</a>, <a href="https://coffeebeans.io/">CoffeeBeans</a></li>
</ul>
<img src="https://media.coffeebeans.io/blogs/blog-tn2.png" alt="cover">
<img src="https://media.coffeebeans.io/blogs/0_olcrXmbB3FH7X2_Z.png" alt="traditional ML vs Transfer learning">
Source: Educative.io
Key Motivation for Transfer Learning– If we look at tasks in our everyday life, we don’t necessarily learn everything from scratch. We essentially try to make our understanding of some new task based on our experiences in the past. It’s also fairly difficult to gather a huge amount of data to train a Deep Learning model from scratch. Transfer learning comes very handy in cases where you don’t have much data to train a new model from scratch.
This article is solely focused on the practical implementation of Transfer Learning by using a pre-trained model and we will further train it according to our requirements.
Let’s walk through the script:
We will start by installing pip, sentencepiece, datasets, and transformers.
<img src="https://media.coffeebeans.io/blogs/0_-JS5IJ4Qctw7r-xq.png" alt="code block">
Next, we will define a class where we would write the function for encoding data, in our case, it’s tweets.
For more details, you can read about the different attributes of tokenizer.encode_plus in the following <a href="https://huggingface.co/transformers/v2.11.0/glossary.html">link</a>.
Also, you can go through this <a href="https://huggingface.co/docs/transformers/main_classes/tokenizer">link</a> as well.
<img src="https://media.coffeebeans.io/blogs/0_p_SesQeHh-ez1GSK.png" alt="code block">
The next steps involve defining hyperparameters as well as the model on which we are going to do training.
There is no strictness regarding the model used here, you can choose your own depending on the problem statement. But it’s always fair to check on its performance before you further train it on your custom data.
<img src="https://media.coffeebeans.io/blogs/0_RIIZ2NbdWuu28tap.png" alt="code block">
Next, we should define a relabeling function that actually defines the labels of your classification problem. Here I am taking three categories 0,1 and 2.
<img src="https://media.coffeebeans.io/blogs/0_JjaBh1lHV0gedtjU.png" alt="code block">
Next is converting the data into an appropriate format before we pass it to a trainer.
<img src="https://media.coffeebeans.io/blogs/0_g1gBbEvTZfBDhulu.png" alt="code block">
<img src="https://media.coffeebeans.io/blogs/0_LtHo6TeA0SgBbXYe.png" alt="code block">
The next step is model training and this is the most crucial step of whole process where we will pass the data to pretrained model trainer.
<img src="https://media.coffeebeans.io/blogs/0_r2_k9DiGp_7KANyw.png" alt="code block">
<img src="https://media.coffeebeans.io/blogs/0_uZsRr_lZw7wQRyku.png" alt="code block">
Please see that I have kept epochs to 10 and the Batch size is 8, my model took approx 3hrs for 1 epoch to train since my dataset was large, you can play around here by keeping epochs less or keeping the dataset small, but if you want to train a good model, my suggestion is to use at least8 to 10 epochs for satisfactory results.
Also, make sure to evaluate your results.
<img src="https://media.coffeebeans.io/blogs/0_E778b0GgLXoPMgyR.png" alt="code block">
<h1>The next concept I want to share is ONNX model format.</h1>
ONNX is an intermediary machine learning framework used to convert different machine learning frameworks. In Short, train your custom model in whatever framework you are comfortable with, like Tensorflow and Pytorch, and just finally convert it to common ONNX format. This provides a lot of flexibility to developers as well. You can also use ONNX models with optimized inference frameworks (Onnx runtime, Ncnn, Tvm etc) to maximize performance.
(I will write a separate blog later for ONNX in detail)
According to official documentation :
<img src="https://media.coffeebeans.io/blogs/0_jILR_D1L-6qfsavy.png" alt="Desctiption of ONNX">
<h1>The next concept I want to share is ONNX model format.</h1>
ONNX is an intermediary machine learning framework used to convert different machine learning frameworks. In Short, train your custom model in whatever framework you are comfortable with, like Tensorflow and Pytorch, and just finally convert it to common ONNX format. This provides a lot of flexibility to developers as well. You can also use ONNX models with optimized inference frameworks (Onnx runtime, Ncnn, Tvm etc) to maximize performance.
(I will write a separate blog later for ONNX in detail)
According to official documentation :
<img src="https://media.coffeebeans.io/blogs/0_wkIEzJg18rRoq6of.png" alt="code block">
<img src="https://media.coffeebeans.io/blogs/0_C-vVDnM2HMyXNaV6.png" alt="code block">
Please note here that defining features is really important as this defines for what specific task we are converting our model to ONNX format. The last step involves converting the model into ONNX format and then checking the export or loading the model and then utilizing it in the production or research environment.
Please suggest any changes if some error is there, the error would be rectified immediately.

Transfer Learning works on the idea that a pre-trained model trained in one setting can be utilized in a different setting for a specific task. I will be doing sentiment analysis here by using a pre-trained BERT model using Pre-Trained Model Approach and will be further training it on a custom dataset prepared.

How to use a pre-trained BERT model for classification using Transfer learning

Linking the customer profile with the features and services to come up with a prioritized and targeted product backlog<img src="https://media.coffeebeans.io/blogs/1_0XX5dG-6s6_X8iE59xldAQ%20(1).jpg" alt="cover">A Product Owner is a team role defined in the Scrum Guide as the team member accountable for maximizing the value of the product resulting from the work of the Team. The work of the Product Owner starts with defining the Product goal and the Product Backlog Items (PBIs) based on inputs from key stakeholders across the organization, market data, competitor analysis and many other factors.
The Product Owner is responsible for defining and prioritising the Product Backlog and also ensuring that it is transparent, visible and understood. For a product to be successful, the market, the team and the product features, all play an important role. From a product owner’s perspective, it is important to ensure that the product developed actually fits the customer values and needs. There may be a good target market and a well designed product but if the product does not fit the market then all the analysis and effort can go to waste.
The Value Proposition Canvas is a tool that can help ensure that a product or service is positioned around what the customer values and their needs. The Value Proposition Canvas, a framework to ensure product-market fit was initially developed by Dr. Alexander Osterwalder.<img src="https://media.coffeebeans.io/blogs/0_SW7hw3psPsE0ikuW.png" alt="image">For a product owner, the Value Proposition Canvas can not only help ensure product-market fit but also be a source for well-structured and prioritised user stories. Let us delve deeper into how the value proposition canvas works and how a product owner can use it efficiently.
As shown above, the Value Proposition Canvas has 2 main components:
<ol>
<li>The SQUARE</li>
<li>The CIRCLE</li>
</ol>
The CIRCLE deals with the customer profile and the SQUARE deals with the value proposition. In simpler words, the CIRCLE talks about the customer, and the SQUARE talks about the product and its features.
For a product owner, the CIRCLE denotes the outcomes or consequences, and the SQUARE denotes what makes the outcomes happen.<img src="https://media.coffeebeans.io/blogs/0_o1n3jrEai47Zo17t.png" alt="image">On the right, the canvas profiles the customer by asking: what jobs are they trying to accomplish, what’s hard or unpleasant (pains) about their current way of doing it, and what’s desirable or delightful (gains) in an ideal solution.
On the left, we map out the value proposition by asking: what is the product or service, and how does it create the gains the customer wants and soothe the pains they don’t.
Now that we know what the two things denote, let us look at how to use them. The PO starts by defining the customer jobs (the functional, social, and emotional tasks customers are trying to perform, problems they are trying to solve, and needs they want to satisfy). These customer jobs help us identify the pains and gains that the customer desires. While it may easily be mistaken, the gains are not the opposite of the pains. The pains are the negative experiences of a customer, while the gains are the benefits that the customer expects and needs.
Let us look at the SQUARE. The Value map (SQUARE) also has 3 components: pain relievers, gain creators, and products and services. The products and services show the features that can help relieve the pain or create gains for the customer, in turn providing value to the customer. The pain relievers detail how the products and services can relieve the pain of the customer, and the gain creators detail how the products and services can offer added value to the customer.
We will now see how all of this ties together. The Gain Creators may provide one or more gains and may also contribute to the relief of one or more of the pains. Similarly, pain relievers can solve one or more pains and may lead to one or more gains for the customer.
For a product owner, if the Gain Creators and Pain Relievers can solve the major pain points and provide gains to the users, and the products and services are taking care of the customer jobs, in other words, if the SQUARE and the CIRCLE are seen as fitting properly with each other, then the product or service is said to be positioned around what the customer values and needs. We can then have a good product-market fit.
Below is the sample Value Proposition Canvas for Uber from back when it was getting off the ground:<img src="https://media.coffeebeans.io/blogs/0_dvqgGDyw-JMOO9bX.png" alt="image">Apart from providing insights into the product-market fit, the Value Proposition Map also provides some other interesting insights for the product owner.
Filling the Gap/Enhance your offering
We talked about how the Value Proposition Canvas tells us how our product features are solving pains and resulting in gains. But if we look at it in reverse order, we observe that the canvas also gives us pains and gains that are not resolved using the current feature set. So, that gives us the GAP in our offering.
We can use this to identify the features that we now want to add that will enhance our offering and make our product more attractive to customers.
Prioritisation
We discover that some Gain Creators will be associated with more than one gain or pain after mapping the relationships between the Gains, Pains, Gain Creators, and Pain Relievers. The same goes for pain relievers. Apart from this, there is an individual priority for the gains and pains as well. These priorities combines with the link between the Value Map (SQUARE) and Customer Profile (CIRCLE), feed into the prioritisation of the Product Backlog by telling us which of the features will provide us a better return on investment.
Writing user Stories
A user story is an articulation of the product backlog items. A user story is the smallest unit of work in an agile framework. It’s an end goal, not a feature, expressed from the software user’s perspective.
As per a recent survey at Thoughtworks, the most preferred format of a user story contains 3 elements: As a____, I want to____, So that____
In this format, we first define the user from whose perspective we want to write the story (as a). We also define the task that the said user wants to accomplish (I want to). Finally, we look at the value that the user wants (so that). This is where the canvas comes in. It provides the gains and pains we want to refer to. A Product Owner would want to ensure that the story leads to pain being relieved or gain being created.
I hope this article helps the readers understand the utility of the Value Proposition Canvas when defining product goals and product backlog. It is a valuable tool in the arsenal of a Product Owner.
Thanks for Reading
Credits: Adyot Kumar Srivastava
Source:
<ol>
<li>CSPO workshop by Lizzy Morris</li>
<li><a href="https://www.scrum.org/resources/what-is-a-product-owner">https://www.scrum.org/resources/what-is-a-product-owner</a></li>
<li><a href="https://www.b2binternational.com/research/methods/faq/what-is-the-value-proposition-canvas/#:~:text=The%20Value%20Proposition%20Canvas%20is,the%20customer%20values%20and%20needs.&#x26;text=It%20is%20a%20detailed%20look,customer%20segments%20and%20value%20propositions">https://www.b2binternational.com/research/methods/faq/what-is-the-value-proposition-canvas/</a></li>
<li><a href="https://www.youtube.com/watch?v=kiWkRiynPAo">https://www.youtube.com/watch?v=kiWkRiynPAo</a></li>
<li><a href="https://interaction.net.au/articles/value-proposition-canvas-explained/">https://interaction.net.au/articles/value-proposition-canvas-explained/</a></li>
<li><a href="https://www.atlassian.com/agile/project-management/user-stories">https://www.atlassian.com/agile/project-management/user-stories</a></li>
</ol>

A Product Owner is a team role defined in the Scrum Guide as the team member accountable for maximizing the value of the product resulting from the work of the Team. The work of the Product Owner starts with defining the Product goal and the Product Backlog Items (PBIs) based on inputs from key stakeholders across the organization, market data, competitor analysis and many other factors.

Value Proposition Canvas — An indispensable tool for Product Owners

A product roadmap is a document that outlines the steps that will be taken for a product or solution to be developed over time. Roadmaps are used by product owners to specify future product functionality and the timing of the release of new features. When used in agile development, a roadmap gives the team’s daily work critical context and should be adaptable to changes in the market.<img src="https://media.coffeebeans.io/blogs/1_PqG8vJjrHNrcYfLSwAfI4g.png" alt="cover">What is a Product Roadmap?
A product roadmap is a document that outlines the steps that will be taken for a product or solution to be developed over time. Roadmaps are used by product owners to specify future product functionality and the timing of the release of new features. When used in agile development, a roadmap gives the team’s daily work critical context and should be adaptable to changes in the market.
Nobody knew what product roadmaps should look like until recently, but we all knew something had to change.
To communicate our product plans at older companies, we used pen and paper and a release planner. They were long, complicated, and detailed — after all, they were designed for developers — and they resulted in a jumbled backlog that was difficult to follow.
Today, road mapping is a divisive topic, but one thing is clear: It is necessary. We all know that no one reads the lengthy piece. We’re also learning how effective a roadmap can be when it’s so well-designed that teams use it to make product decisions and businesses use it to make business decisions.
We’ll look at how a roadmap can help you connect your work with that of others and reclaim control over your product. But what exactly does that road map entail? Let’s dive in to understand it better.
What does a great product roadmap look like?
A good product roadmap must be visually appealing, easily accessible, and clear enough for anyone to scan for answers to the following questions:
· What exactly are we doing?
· Why are we doing it in the first place?
· What does this have to do with our OKRs?
OKRs are defined as Objectives and Key Results that the product is entitled to achieve through the set of features you are building for the product. It encompasses the milestones you will attain in releasing those features.
A theme-based product roadmap is based on this premise, and the benefits are enormous and immediate.
· Reduce the number of meetings you attend — On the roadmap, your priorities (what and why) are clearly stated. You do not need to explain things to different people in different ways.
· Encourage constructive team debates — Your roadmap can serve as a benchmark for team members to hold themselves and one another accountable for linking deliverables to roadmap goals and OKRs.
· Make product choices, everyone understands — you’re no longer the jerk who dismisses ideas. You can actually discuss customer feedback and ideas using your roadmap and priorities, which are visible to everyone
We must abandon the notion that we can enumerate a set of features that will represent what we will do in the future. This is a ridiculous notion. We should communicate how we will make decisions instead of sharing feature lists with the rest of the company.
A theme-based roadmap is made to be adaptable, so your plans can change as your circumstances change, and it can serve as a communication tool.
o begin, make three columns.
A roadmap, which includes release dates, may be familiar to you. A theme-based roadmap substitute with time horizons made up of three columns:
1. Now — Tasks that you are currently working on
2. Next — Tasks that are coming up soon
3. Later — Tasks that you’d like to work on in the future or need to do a bit more research before you move on
It’s worth noting that we’re not displaying any dates. This isn’t a release schedule; rather, it’s a bird’s eye view of your top priorities. Those are always subject to change, especially in the far future, which you can’t plan for today meaningfully.
The idea is to give yourself some ﬂexibility in case things change. You can push something back if it was current but you no longer want it.
Organize Broad Themes
A roadmap, which includes a list of features and their expected completion date, may be familiar to us.
A theme-based map replaces this with themes, which appear as cards on your roadmap.
Those words come from the realization that almost any product will receive a flood of customer feedback and feature requests that revolve around a single universal problem.
You can either become a slave to those requests or take a step back and address the root of the problem with a single, elegant solution.
Working in themes allows you to think about how you might approach each problem and then decide how to solve it later.
<img src="https://media.coffeebeans.io/1_fCW9nhdBAAmBTa9mEPdYrw.png" alt="image">In user administration in the blue cloud, you can observe it’s considered as them because it consists of certain features that complete that. Sign up for a new account and login for existing users are two basic features that get covered under it. It’s basically a strategic theme to add/login users to an account, not a specific deliverable.
But on a contrary, you can observe that the last one is crossed out. Because you want your cards to be strategic. To be a strategy, “rewriting transactional emails” is far too specific. It’s both an idea and a tactic, and does not belong in the high-level view you’re developing in this step.
It would, however, be more appropriate as an idea within a larger-scale roadmap card.
Inserting Supporting Details
Once you’ve nailed down your themes, you can add more supporting information for anyone who wants to dig deeper. These details help us strengthen what we’re putting on our roadmap, which could include information that’s useful to those who read it:
1. What exactly are we doing?
2. Why are we doing it in the first place?
3. What does this have to do with our OKRs?
OKRs are defined as Objectives and Key Results that the product is entitled to achieve through the set of features you are building for the product. It encompasses the milestones you will attain in releasing those features.
Internally, your team will have access to detailed data that will assist them in preparing for and navigating your workflow. The following are some of the details:
· Ideas — Strategic suggestions for improvement. These concepts respond to a straightforward business question: What problem are you trying to solve?
· Customer feedback — We attach feedback to ideas in Product Board so that it can be linked to potential improvements and we can keep track of what our customers want.
· User stories — For inspiration, consider the following case scenarios: I want to X in order to Y as a user.
Cards in the Future column don’t have to have all of those answers just yet, but they should get a lot more detailed as they get closer to the ‘Current’ term.
Label and Categorize
You can color code and tag your roadmap as you go along to make it easier for the viewer to sort through and filter down based on a specific interest. It’s similar to how you use post-it notes!
Remember, color coding can also help to distinguish certain features based on the parameters you choose like prioritization, efforts, immediate release, customer needs, etc. It simplifies your process of developing features and functionalities for your product.
<img src="https://media.coffeebeans.io/blogs/1_AX2f8QEvpegMdAc6howlyA.png" alt="image">User Profiles, maps of cuisines, and reviews are something most customers look for before going to any restaurant. Similarly, what needs to be built when and for whom, the prioritization should be customer-centric rather than features-centric.
Similarly, you can find next and later features to be developed and in case you want to make changes, make sure it helps to solve your customers’ pain points.
How’s that for basic usability? This keeps things simple and engaging visually, and everyone will appreciate not having to slog through stacks of cards to find what they came for.
Rather than staring blankly at one large roadmap, your co-workers can concentrate on the ones that are relevant to them.
The Bottom Line: This Roadmap Is Much Larger Than You Think
The idea’s sheer brilliance is how, with a single gesture, understanding everything about customers becomes the most influential factor in developing product strategy. “Look at us,” companies say when they talk about features. Take a look at what we’ve accomplished thus far.” “Take a look at what you’re dealing with,” companies say when they discuss customer problems. Take a look at what we’re planning to do to facilitate.”
And, in a single stroke, you’re empowering everyone at your company to think in this manner. You are not the only one. Can you imagine the impact that would have?
Just stick to this famous quote by Jeff Bezos:
References
· <a href="https://happyproject.in/stoic-decisions/">https://happyproject.in/stoic-decisions/</a>
· <a href="https://www.quora.com/What-is-the-better-way-to-build-an-effective-product-roadmap-What-makes-a-good-product-roadmap">https://www.quora.com/What-is-the-better-way-to-build-an-effective-product-roadmap-What-makes-a-good-product-roadmap</a>
· <a href="https://www.prodpad.com/resources/roadmap-course/">https://www.prodpad.com/resources/roadmap-course/</a>
· <a href="https://railsware.com/blog/product-roadmap/">https://railsware.com/blog/product-roadmap/</a>
· <a href="https://productschool.com/blog/product-management-2/product-roadmap-templates/">https://productschool.com/blog/product-management-2/product-roadmap-templates/</a>
· <a href="https://www.smartsheet.com/your-complete-guide-understanding-and-creating-visual-product-roadmaps">https://www.smartsheet.com/your-complete-guide-understanding-and-creating-visual-product-roadmaps</a>
By Rahul Kumar

A product roadmap is a document that outlines the steps that will be taken for a product or solution to be developed over time. Roadmaps are used by product owners to specify future product functionality and the timing of the release of new features. When used in agile development, a roadmap gives the team’s daily work critical context and should be adaptable to changes in the market.

How to Create a Product Roadmap That Everyone Can Understand!

<img src="https://media.coffeebeans.io/blogs/1_Rp6x69Ruif29tN7FYDA1Ng.jpg" alt="cover">
So the question is now: What is Prompt Engineering?
Prompt engineering involves deliberately crafting input instructions to steer AI language models, enhancing output precision and control while generating responses.
Here are some Basic and Advance Prompt Engineering Techniques:
<ol>
<li>Zero Shot Prompting</li>
<li>Few Shot Prompting</li>
<li>Chain of Thought Prompting (COT)</li>
<li>Tree of Thought Prompting (TOT)</li>
</ol>
<img src="https://media.coffeebeans.io/blogs/0_TgsfGF2EwppdFnaB.jpg" alt="reference image">
Let’s discuss each Prompting Technique with code examples.
So for that we will use <a href="https://platform.openai.com/overview">OpenAI</a>, <a href="https://python.langchain.com/docs/get_started/introduction.html">Langchain</a> (Open source framework for development of LLM based applications).
Initial setup code :
<pre><code class="language-python">import openai
import os
from langchain.llms import OpenAI
from langchain.agents import load_tools
from langchain.agents import initialize_agent

from dotenv import load_dotenv, find_dotenv
_ = load_dotenv(find_dotenv())

openai.api_key = os.getenv('OPENAI_API_KEY')
def get_completion(prompt, model="gpt-3.5-turbo"):
 messages = [{"role": "user", "content": prompt}]
 response = openai.ChatCompletion.create(
 model=model,
 messages=messages,
 temperature=0, # this is the degree of randomness of the model's output
 )
 return response.choices[0].message["content"]
</code></pre>
We need to Install openai, langchain and dotenv. We can use the pip command for that. And for OPEN_API_KEY create an account on openai, generate the key and then set it in your environment. Command to set up the OPEN_API_KEY in the environment.
<pre><code class="language-python">export OPEN_API_KEY=sk..........
</code></pre>
Let’s begin:
<ol>
<li>Zero Shot Prompting:</li>
</ol>
Zero shot prompting enables a model to make predictions about previously unseen data without the need for any additional training. So LLM models have already trained on very large data sets, that’s why for some simpler queries they can predict the result easily.
Example:
<pre><code class="language-python">prompt = """
Classify the text into neutral, negative or positive. 
Text: I think the vacation is okay.
Sentiment:
"""
response = get_completion(prompt)
print(response)

Output: Neutral
</code></pre>
So in the above example, LLM model simply predicts the sentiment, because the model has that much knowledge.
2. Few Shot Prompting:
Despite impressive zero-shot abilities, large language models struggle with complex tasks in that mode. Few-shot prompting, using prompts with examples, aids in-context learning and enhances model performance.
Example:
<pre><code class="language-python">prompt = """
A "Lion" is a Big, dangerous animal native to Africa. 
An example of a sentence that uses the word Lion is: We were travelling in Africa and we saw these very dangerous Lions. 
To do a "enthusiasm" means to jump up and down really fast. An example of a sentence that uses the word enthusiasm is:
"""
response = get_completion(prompt)
print(response)

Output: The crowd erupted with enthusiasm as their favorite team scored the winning goal.
</code></pre>
Drawback:
Standard few-shot prompting works well for many tasks but is still not a perfect technique, especially when dealing with more complex reasoning tasks.
Example:
<pre><code class="language-javascript">prompt = """
The odd numbers in this group add up to an even number: 4, 8, 9, 15, 12, 2, 1.
A: The answer is False.
The odd numbers in this group add up to an even number: 17, 10, 19, 4, 8, 12, 24.
A: The answer is True.
The odd numbers in this group add up to an even number: 16, 11, 14, 4, 8, 13, 24.
A: The answer is True.
The odd numbers in this group add up to an even number: 15, 32, 5, 13, 82, 7, 2.
A:
"""
response = get_completion(prompt)
print(response)

Output: The answer is False.
</code></pre>
Here we know that the answer is (15+5+13+7) = 40, which is an even number, But our model is not able to generate it. But we can resolve this problem by using the advanced prompt engineering technique COT Prompting.
3. Chain of Thought Prompting:
Chain of Thought (COT) prompting enables complex reasoning capabilities through intermediate reasoning steps. It helps in LLM models to understand the problem in a more detailed way. Let’s take the same example above. But we will provide some additional reasoning intermediate steps.
Example:
<pre><code class="language-python">prompt = """
The odd numbers in this group add up to an even number: 4, 8, 9, 15, 12, 2, 1.
A: Adding all the odd numbers (9, 15, 1) gives 25. The answer is False.
The odd numbers in this group add up to an even number: 17, 10, 19, 4, 8, 12, 24.
A: Adding all the odd numbers (17, 19) gives 36. The answer is True..
The odd numbers in this group add up to an even number: 15, 32, 5, 13, 82, 7, 2. 
A:
"""
response = get_completion(prompt)
print(response)

Output: A: Adding all the odd numbers (15, 5, 13, 7) gives 40. The answer is True.
</code></pre>
No, we got the right answer, because the LLM model tries to solve the question with steps.
Take one more example:
<pre><code class="language-python"># first ask the answer directly:
prompt = """
please solve (10 + 2) * (-4 - 1), give direct answer
"""
response = get_completion(prompt)
print(response)

#we get wrong answer
Output: The direct answer to (10 + 2) * (-4 - 1) is -72. 
# ask the answer in steps:
prompt = """
please solve (10 + 2) * (-4 - 1), please solve in steps
"""
response = get_completion(prompt)
print(response)

Output:

Step 1: Simplify the expression inside the parentheses.
(10 + 2) * (-4 - 1) = 12 * (-4 - 1)

Step 2: Simplify the expression inside the parentheses.
12 * (-4 - 1) = 12 * (-5)

Step 3: Multiply the two numbers.
12 * (-5) = -60

Therefore, (10 + 2) * (-4 - 1) = -60.

#this is the right answer
</code></pre>
Here we just saw COT is a powerful Prompt engineering technique, If we are passing proper intermediate steps.
4. Tree-of-Thought Prompting :
ToT prompting gives superior reasoning abilities to LLM models, This Tree-of-Thought Prompting technique permits Large Language Models to rectify their errors autonomously while progressively accumulating knowledge.
Let’s try to understand this via example:
First try to solve this with zero-shot prompting.
<pre><code class="language-python">prompt = """
Bob is in the living room.
He walks to the kitchen, carrying a cup.
He puts a ball in the cup and carries the cup to the bedroom.
Then he walks to the study room and
He turns the cup upside down, then walks to the garden.
He puts the cup down in the garden, then walks to the garage.
Where is the ball?
"""
response = get_completion(prompt)
print(response)

Output: The ball is in the cup in the garden.
</code></pre>
Right Answer: The ball is in the study room. But, We got the wrong answer, Now we will try to solve this via COT prompting. Here is the
Example:
<pre><code class="language-python">prompt = """
Bob is in the living room.
He walks to the kitchen, carrying a cup.
He puts a ball in the cup and carries the cup to the bedroom.
Then he walks to the study room and
He turns the cup upside down, then walks to the garden.
He puts the cup down in the garden, then walks to the garage.
Where is the ball?
Think carefully and logically, explaining your answer.
"""
response = get_completion(prompt)
print(response)

Output: The ball is in the living room.
</code></pre>
Still, we are not getting the right answer. Let’s try to solve it with TOT Prompting. The principle is to provide the LLM an opportunity to explore multiple branches of reasoning while self-evaluating as it progresses.
<pre><code class="language-python">prompt = """
Imagine three different experts are answering this question.
All experts will write down 1 step of their thinking,
then share it with the group.
Then all experts will go on to the next step, etc.
If any expert realises they're wrong at any point then they leave.
The question is in sigle quotes:
'Bob is in the living room.
He walks to the kitchen, carrying a cup.
He puts a ball in the cup and carries the cup to the bedroom.
Then he walks to the study room and
He turns the cup upside down, then walks to the garden.
He puts the cup down in the garden, then walks to the garage.
Where is the ball?'
"""
response = get_completion(prompt)
print(response)

Output:
Expert 1: "Based on the information given, Bob starts in the living room and walks to the kitchen while carrying a cup."
Expert 2: "Next, Bob puts a ball in the cup and carries the cup to the bedroom."
Expert 3: "Then, Bob walks to the study room and turns the cup upside down."
Expert 1: "After that, Bob walks to the garden and puts the cup down there."
Expert 2: "Finally, Bob walks to the garage."
Expert 3: "Since the ball was put in the cup and the cup was turned upside down in the study room, the ball should still be in the study room. Therefore, the ball is not in the garage."
</code></pre>
Now we got the right answer. Because LLM creates multiple branches of reasoning and in-progress self-evaluation to solve the problem, And finally merge all the branches together to get the response.
Conclusion:
There are a couple of more powerful prompt engineering techniques like ReAct Prompting, Generated Knowledge Prompting, Self-Consistency, and so on, That we can use while creating a good prompt. We can choose these techniques on the basis of our use cases. Sometimes a basic technique can work instead of the advanced one.
Note: As all we know Prompt Engineering is a topic of research so most of the examples are based on research papers. While LLM models are evolving and in this article, we used model got-3.5-turbo and if you use any other model you might get some different results that is totally ok, because this article is based on how we can write more powerful prompts that solve our business problems.

Today we are all in the world of Generative AI, And we all are using some of the powerful GEN AI tools like ChatGPT, DALL-E, and GitHub Copilot. But the most important part while working with these tools is, How efficient you are in Prompt Engineering.

What developers need to know about Prompt Engineering

Serverless computing has been gaining popularity as a powerful and efficient solution for building and deploying applications. With serverless computing, developers can focus on writing code without worrying about managing servers or infrastructure. In this blog post, we’ll take a deep dive into the architecture and capabilities of serverless computing, exploring how it works, its benefits and drawbacks, and its potential applications.<img src="https://media.coffeebeans.io/blogs/Blog%20cb.jpeg" alt="cover">
<h1>What is Serverless Computing?</h1>
Serverless computing is a cloud computing model where the cloud provider manages the infrastructure and automatically provisions resources as needed to execute code. This means that developers don’t have to worry about managing servers, scaling, or infrastructure maintenance. Instead, they can focus on writing code and building applications. Serverless computing is often used for building event-driven applications or microservices, where functions are triggered by events and execute specific tasks.
<h1>How Serverless Computing Works</h1>
In serverless computing, applications are broken down into small, independent functions that are triggered by specific events. These functions are stateless, meaning they don’t retain information between executions. When an event occurs, the cloud provider automatically provisions the necessary resources and executes the function. Once the function is complete, the resources are de-provisioned, making serverless computing highly scalable and cost-efficient.
<h1>Serverless Computing Architecture</h1>
The architecture of serverless computing typically involves four components: the client, the API Gateway, the compute service, and the data store. The client sends requests to the API Gateway, which acts as a front-end to the compute service. The compute service executes the functions in response to events and may interact with the data store to retrieve or store data. The API Gateway then returns the results to the client.
<h1>Benefits of Serverless Computing</h1>
Serverless computing offers several benefits over traditional server-based computing, including:
Reduced costs: Serverless computing allows organizations to pay only for the resources they use, rather than paying for dedicated servers or infrastructure.
Improved scalability: Serverless computing can automatically scale up or down depending on demand, making it highly scalable and efficient.
Reduced maintenance: Since the cloud provider manages the infrastructure, organizations don’t need to worry about maintaining servers or infrastructure.
Faster time to market: Serverless computing allows developers to focus on writing code and building applications, reducing the time to market new products and services.
<h1>Drawbacks of Serverless Computing</h1>
While serverless computing has several benefits, it also has some drawbacks, including:
Limited control: Since the cloud provider manages the infrastructure, developers have limited control over the environment and resources.
Cold start times: When a function is executed for the first time, it may take longer to start up, leading to slower response times.
Vendor lock-in: Organizations may be tied to a specific cloud provider, making it difficult to switch providers or migrate to a different environment.
<h1>Some facts about serverless computing</h1>
Serverless computing is often referred to as Functions-as-a-Service (FaaS) because it allows developers to write and deploy individual functions rather than entire applications.
Serverless computing is often used in microservices architectures, where applications are broken down into smaller, independent components that can be developed, deployed, and scaled independently.
Serverless computing can result in significant cost savings for organizations because they only pay for the resources they use. This can be especially beneficial for applications with unpredictable traffic patterns or occasional bursts of computing power.
One of the biggest drawbacks of serverless computing is the “cold start” problem, where a function may take several seconds to start up if it hasn’t been used recently. However, this problem can be mitigated through various optimization techniques.
Serverless computing is often used in event-driven architectures, where functions are triggered by specific events such as user interactions, changes to a database, or changes to a file system. This can make it easier to build highly scalable and efficient applications.
Now, let’s explore some other serverless computing frameworks that can be used in addition to Google Cloud Functions.
AWS Lambda: AWS Lambda is a serverless compute service from Amazon Web Services (AWS). It allows developers to run code in response to events without worrying about managing servers or infrastructure.
Microsoft Azure Functions: Microsoft Azure Functions is a serverless compute service from Microsoft Azure. It allows developers to run code in response to events and supports a wide range of programming languages.
IBM Cloud Functions: IBM Cloud Functions is a serverless compute service from IBM Cloud. It allows developers to run code in response to events and supports a wide range of programming languages.
OpenFaaS: OpenFaaS is an open-source serverless framework that allows developers to run functions on any cloud or on-premises infrastructure.
Apache OpenWhisk: Apache OpenWhisk is an open-source serverless platform that allows developers to run functions in response to events. It supports a wide range of programming languages and can be deployed on any cloud or on-premises infrastructure.
Kubeless: Kubeless is a Kubernetes-native serverless framework that allows developers to run functions on Kubernetes clusters. It supports a wide range of programming languages and can be deployed on any Kubernetes cluster.
IronFunctions: IronFunctions is an open-source serverless platform that allows developers to run functions on any cloud or on-premises infrastructure. It supports a wide range of programming languages and can be deployed on any container orchestrator.
These serverless computing frameworks offer developers a range of options for building and deploying serverless applications. Each framework has its own strengths and weaknesses, so developers should choose the one that best fits their needs.
<h1>Real-time examples</h1>
Coca-Cola: Coca-Cola uses serverless computing to power its Freestyle soda machines, which allow customers to mix and match different soda flavors. The machines use AWS Lambda functions to process customer requests and make recommendations based on their preferences.
iRobot: iRobot uses serverless computing to power its Roomba robot vacuums, which use computer vision and machine learning to navigate homes and clean floors. The Roomba vacuums use AWS Lambda functions to process data from their sensors and decide where to go next.
Capital One: Capital One uses serverless computing to power its mobile banking app, which allows customers to manage their accounts, transfer money, and pay bills. The app uses AWS Lambda functions to process requests and deliver real-time information to users.
Fender: Fender uses serverless computing to power its Fender Play platform, which provides online guitar lessons to users around the world. The platform uses AWS Lambda functions to process user data and generate personalized lesson plans.
Netflix: Netflix uses serverless computing to power its video encoding and transcoding workflows, which are used to prepare video content for streaming on various devices. The workflows use AWS Lambda functions to process video files and convert them into the appropriate format for each device.
Conclusion
Serverless computing is a powerful and efficient solution for building and deploying applications. It offers several benefits, including reduced costs, improved scalability, reduced maintenance, and faster time to market. However, it also has some drawbacks, including limited control, cold start times, and vendor lock-in. Despite these drawbacks, serverless computing will likely become an increasingly popular solution for building event-driven applications and microservices.

Serverless computing has been gaining popularity as a powerful and efficient solution for building and deploying applications. With serverless computing, developers can focus on writing code without worrying about managing servers or infrastructure. In this blog post, we’ll take a deep dive into the architecture and capabilities of serverless computing, exploring how it works, its benefits and drawbacks, and its potential applications.

When to use serverless computing

<img src="https://media.coffeebeans.io/blogs/Untitled-3-12-2048x1152.jpg" alt="cover">
Architectural Differences: 
The fundamental difference between ARM and x86 processors lies in their architectural designs. ARM processors have a simplified instruction set, with each instruction completed in one cycle. On the other hand, x86 processors have a more complex instruction set, offering a wide range of instructions, some of which may require multiple cycles to execute. While ARM processors execute a finite number of commands, x86 processors have a much larger instruction set, with approximately 981 instructions in x86-64 architecture. This difference in architecture affects the execution efficiency, programming complexity, and overall performance of the processors.
RISC vs CISC and Development Impact:
The RISC architecture of ARM processors emphasizes efficiency in cycles per instruction, while the CISC architecture of x86 processors focuses on efficiency in instructions per program. RISC processors execute simpler instructions within a cycle at a high clock speed, while CISC processors can perform complex operations with a single instruction. This distinction impacts development as RISC machines may require multiple instructions to execute tasks that can be done with a single CISC instruction. Both architectures have their own advantages and trade-offs, and the choice depends on the specific application and development requirements.
Energy Usage and Efficiency:
ARM processors are designed with a focus on low power consumption and energy efficiency. Their integration into SoCs allows for overall resource management, resulting in lower heat production and simplified cooling systems. This makes ARM processors ideal for mobile devices like smartphones, where battery life is crucial. In contrast, x86 processors prioritize high processing speed and performance, which often results in higher power consumption and heat generation. While x86 processors excel in raw computing power, they tend to consume more energy compared to ARM processors.
Market Trends and Adoption:
ARM processors have gained significant traction in the PC market in recent years. Apple’s transition to ARM-based M-series processors in their Macs has been a game-changer, with Apple accounting for around 90% of ARM-based PCs. ARM processors also power a significant number of inexpensive Chromebooks. Market projections suggest that ARM’s market share in the PC market could grow from the current 14% to 25% by 2027. The availability of more native Arm-based applications and the expertise of existing PC OEMs/ODMs and smartphone manufacturers entering the market contribute to the growing popularity of ARM-based PCs.
<ol>
<li>Advantages and Disadvantages of ARM Processors:</li>
</ol>
<ul>
<li>Advantages:
<ul>
<li>Lower costs: ARM processors are affordable to create and manufacture.</li>
<li>Simple design: ARM processors have a simpler architecture, making them easier to understand and implement.</li>
<li>Low power requirements: ARM processors consume less power, resulting in extended battery life for mobile devices.</li>
<li>Reduced heat generation: ARM processors produce less heat, reducing the need for complex cooling solutions.</li>
<li>System-on-a-Chip (SoC) integration: ARM processors can be integrated into a single chip, resulting in compact designs and efficient resource management.</li>
</ul>
</li>
<li>Disadvantages:
<ul>
<li>Compatibility: ARM processors face challenges with compatibility with x86-based software, such as the Windows operating system. However, this is changing with the availability of ARM-compatible versions of popular software.</li>
<li>Limited calculation capacity: Some ARM processors may have limitations in terms of raw computing power, making them less suitable for resource-intensive tasks.</li>
<li>Dependence on skilled programmers: Due to the simplified instruction set of ARM processors, software optimization and skilled programming are essential to achieve maximum performance.</li>
</ul>
</li>
</ul>
<ol>
<li>Advantages and Disadvantages of x86 Processors:</li>
</ol>
<ul>
<li>Advantages:
<ul>
<li>High computing power: x86 processors are known for their high performance and computing power, making them suitable for resource-intensive applications and tasks.</li>
<li>Wide software compatibility: x86 processors have been in the market for a long time, resulting in a vast library of software and a wide range of compatible applications.</li>
<li>Ease of programming: x86 processors offer a more straightforward programming experience due to their complex instruction set, enabling developers to write efficient code with fewer instructions.</li>
</ul>
</li>
<li>Disadvantages:
<ul>
<li>Higher power consumption: x86 processors consume more power compared to ARM processors, which can result in increased energy costs and shorter battery life in mobile devices.</li>
<li>Heat generation: Due to their high power consumption, x86 processors tend to generate more heat, requiring robust cooling solutions.</li>
<li>Potential cost: x86 processors can be more expensive, especially for high-end models, making them less accessible for budget-conscious consumers.</li>
</ul>
</li>
</ul>
Endnote:
Choosing between ARM and x86 architectures depends on various factors, including the target application, performance requirements, power efficiency needs, and software compatibility. ARM processors excel in power efficiency, compact designs, and mobile applications, while x86 processors offer higher computing power, wider software compatibility, and versatility for PCs and servers. As the market evolves, ARM-based PCs are gaining momentum, driven by advancements in ARM-based hardware, software compatibility improvements, and the convergence of laptop and tablet form factors. It is important to weigh the advantages and disadvantages of each architecture when making decisions about hardware and software compatibility, ensuring that your choice aligns with your specific requirements and priorities.

When it comes to computer processors, two prominent architectures stand out: ARM and x86. While both have their strengths and weaknesses, they are fundamentally different in terms of design and application. ARM processors are based on the RISC (Reduced Instruction Set Computing) architecture, while x86 processors follow the CISC (Complex Instruction Set Computing) architecture. In this blog post, we will delve into the comparison between ARM and x86 architectures, exploring their core differences, impact on development, energy usage, advantages, and disadvantages

ARM vs. x86: Comparing Architectures and Their Impact

Architecture design is a must while developing a SaaS application to ensure its scalability and optimising infrastructure costs. In this blog, Lets discuss the implementation of one such architecture with Quarkus java framework and Hibernate ORM
By <a href="https://www.linkedin.com/in/vishal-bihani-821146148/">Vishal Bihani</a> , <a href="https://coffeebeans.io/">CoffeeBeans Consulting</a>
<img src="https://media.coffeebeans.io/blogs/1_d12p7IPbrPHDTNI_YZeFBg.jpg" alt="cover">
Suppose you have a great product idea, you started working on it and developed a cool SaaS product. You get new <a href="https://en.wikipedia.org/wiki/Multitenancy">tenants</a> to use your product and they liked it. In a nutshell, your product is a hit.
<div class="tenor-gif-embed" data-postid="13865195" data-share-method="host" data-aspect-ratio="1.77515" data-width="100%"><a href="https://tenor.com/view/swimming-in-money-rich-laughing-hysterically-money-dive-jaime-lorente-gif-13865195">Swimming In Money Rich GIF</a>from <a href="https://tenor.com/search/swimming+in+money-gifs">Swimming In Money GIFs</a></div> <script type="text/javascript" async src="https://tenor.com/embed.js"></script>
Everything seems to be going well until you realize you are bleeding money with every increasing tenant. You want to scale up but you are unable to do so because you have already exhausted all your cash.
<div class="tenor-gif-embed" data-postid="16909009" data-share-method="host" data-aspect-ratio="1.33333" data-width="100%"><a href="https://tenor.com/view/broke-money-cash-ugh-bart-gif-16909009">Broke Money GIF</a>from <a href="https://tenor.com/search/broke-gifs">Broke GIFs</a></div> <script type="text/javascript" async src="https://tenor.com/embed.js"></script>
You try to figure out what is going wrong and eventually look into your application infrastructure costs.
You understood that every time a new tenant onboards your application you deploy a new server, database, etc. for that tenant. Separate servers and/or databases are only useful for computing or data-intensive applications but yours is neither of that.
In such scenarios, multi-tenant architecture can be very useful in designing applications with shared resources. Multiple combinations are possible some of which are:
1. Separate servers and shared database
2. Shared server and separate databases
3. Shared server and shared database
<h1>How should I decide which resource to keep separate and which to be shared?</h1>
The decision should be made on each resource after looking at its constraints for scaling up. It’s good if each of your resources scales individually based on the utilization rather than complete application scaling up and remaining underutilized.
There are a few things one should keep in mind while deciding.
For servers:
If your business logic is very customized for each tenant then shared servers are not an optimal solution.
OR
If your application is compute-intensive or each tenant is limited by their respective compute then separate servers are a good solution.
For databases:
Generally, it’s a trade-off between data isolation and resource utilization. Development complexity and scaling constraints are also key decision factors.
A separate database provides tenant-specific customization freedom and ease in monitoring per tenant usage and billing them for backup and disaster recovery.
<h1>Multi-tenant database patterns</h1>
There are three types of patterns:
1. Database per tenant
2. Schema per tenant
3. Row per tenant
Database per tenant
Separated databases will be provisioned for each tenant. It achieves maximum data isolation and will give you freedom for high customization and tenant-specific operations to the database.
The development complexity and cost of this approach are high.
Schema per tenant
Separate schemas are created for each tenant. It achieves data isolation at schema level.
High customization and tenant-specific operations like backup, monitoring, and encryption are not supported in many databases.
The development complexity is high but the cost is low.
Row per tenant
In this approach, all the tenants’ data are stored in the same table. Each row is identified by the tenant identifier. Some databases like PostgreSQL support RLS (Row-Level Security) where the rows are implicitly filtered based on the tenant and there is no need to explicitly write the WHERE clause with the tenant identifier.
There is no data isolation and more care has to be put into writing database queries at the application level.
The development complexity and cost are low.
In this blog, we will discuss the implementation of the shared server and schema-based shared database architecture.
<h1>What we will build?</h1>
We will build an app that will store the information of the users of the tenants.
Each tenant will have its separate schema which will store its users’ details. A public schema that will be accessed by every tenant will contain the pricing detail of the product.
Database pattern
<img src="https://media.coffeebeans.io/blogs/0_VF4dTQkulyUgPx7S.png" alt="data base">
schema-based multi-tenancy database pattern
Tables
<img src="https://media.coffeebeans.io/blogs/0_pxSfBVA6FaifGRS0.png" alt="tables schema">
table details
<h1>Implementation</h1>
One important factor while approaching this solution is whether or not frameworks or tools are available that support this architecture and pattern.
For the implementation part, we will be using the Quarkus framework to deploy the REST APIs and perform the business logic.
The Quarkus is an open-source java framework developed by Red Hat which empowers the developers to build cloud-native and modern java applications. You can learn more about Quarkus <a href="https://quarkus.io/">here</a>.
PostgreSQL will be used as the database. It is an open-source SQL database.
Hibernate-ORM supports schema-based multi-tenancy.
Let’s start
Go to <a href="https://code.quarkus.io/">https://code.quarkus.io/</a> and fill in the application configuration.
<img src="https://media.coffeebeans.io/blogs/0_YRbAMlWwbZm4coUl.png" alt="configure application">
Application configuration
Add the following extensions
<img src="https://media.coffeebeans.io/blogs/0_lHGNJmSSv7_Ct0Gp.png" alt="generate application">
Selected extensions
Open in an IDE and run the following command from the terminal:
<pre><code class="language-shell">mvn compile quarkus:dev
</code></pre>
It will download all the dependencies and start the server with the dev profile activated. In dev, profile Quarkus provides the support for live reloading.
After the server has started, perform a GET request to the following URL:
<a href="http://localhost:8080/hello">http://localhost:8080/hello</a>
It should return ‘Hello RESTEasy’
Create entities
Package information is important, all the entities of the public (shared) schema should reside in the same package. Similarly for the tenant schema. We will discuss this in detail later while discussing multiple persistence units.
<a href="https://medium.com/media/ef34cf9db5e75b5fc6c7b508d068bd1e">https://medium.com/media/ef34cf9db5e75b5fc6c7b508d068bd1e</a>
Hibernate does not support database generation (create or update) in the multi-tenant mode so we need to create the databases ourselves.
Create pricing_details table on public schema
<pre><code class="language-json">CREATE TABLE public.pricing_details
(
 id integer,
 product_name character varying NOT NULL,
 price integer NOT NULL,
 created_at date,
 updated_at date,
 PRIMARY KEY (id)
);
Create schema for Tenant A (GOOGL) and Tenant B (AAPL)

CREATE SCHEMA googl
 AUTHORIZATION postgres;

CREATE SCHEMA aapl
 AUTHORIZATION postgres;
Create user_details table on both schema

CREATE TABLE googl.user_details
(
 id integer,
 first_name character varying NOT NULL,
 last_name character varying NOT NULL,
 created_at date,
 updated_at date,
 PRIMARY KEY (id)
);CREATE TABLE aapl.user_details
(
 id integer,
 first_name character varying NOT NULL,
 last_name character varying NOT NULL,
 created_at date,
 updated_at date,
 PRIMARY KEY (id)
);
</code></pre>
Insert values in the tables.
Create repositories for entities
<a href="https://medium.com/media/b7ad2829e2f9a60984a3d216724202c5">https://medium.com/media/b7ad2829e2f9a60984a3d216724202c5</a>
Configuring application properties
write the following configuration properties in the application.properties
<a href="https://medium.com/media/f8ff107dfcfa6acdebc505b21b946d81">https://medium.com/media/f8ff107dfcfa6acdebc505b21b946d81</a>
Only a single data source is configured because we have only one database to connect to.
A persistence unit defines a set of all entity classes that are managed by EntityManager instances in an application
We have separate persistence units for public schema (&#x3C;default>) and tenant-specific schema (tenant)
The data source is set to &#x3C;default> so that they connect to the default data source. Multi-tenant mode is set to SCHEMA
But we need to configure which persistence unit to manage which entities. For that, we need to create package-info.java files in the packages containing the entity files of public and tenant-specific files.
<a href="https://medium.com/media/2ee973e87d56c5fec30db9633af8fa78Inside">https://medium.com/media/2ee973e87d56c5fec30db9633af8fa78Inside</a> the package containing the entities of the ‘public’ schema
<a href="https://medium.com/media/89a6ee2e71ad4bb1c359c7e059b38b6bInside">https://medium.com/media/89a6ee2e71ad4bb1c359c7e059b38b6bInside</a> the package containing the entities of tenant-specific schema
Okay, we configured which persistence should handle which entities but what about schema information?
Each tenant will have its own schema name then how will the persistence unit know which schema to go for. For this, we will create a Tenant Resolver which will resolve the schema name based on the incoming request.
We need to create a Tenant Resolver for all the persistence units.
We will implement the hibernate’s TenantResolver interface which provides a method to resolve schema names.
<a href="https://medium.com/media/c0c771f1f279bde9dc869a368e7883e9">https://medium.com/media/c0c771f1f279bde9dc869a368e7883e9</a>
The @PersistenceUnitExtension takes the name of the persistence unit for which this TenantResolver implementation is.
@PersistenceUnitExtension is for the default persistence unit.
The @RequestScoped is for defining the lifecycle of this bean same to that of the request the application receives (Bean is created when the application receives the request and is destroyed when the request is completed). Once the request is completed it will be destroyed and will be recreated with every incoming request.
what about the below code?
<pre><code class="language-java">public String resolveTenantId() {
String tenantName = context.request().getHeader("tenant");
 return tenantName;
}
</code></pre>
To keep the implementation simpler we will pass the tenant identifier as the header in the HTTP request with the key as “tenant” and the value as the identifier. The above code will retrieve that value and return it as the schema name for that tenant.
Now all the configuration is done but where are the APIs for fetching data?
<a href="https://medium.com/media/bb3dc31fe4641c24aefcab1dd47d3b1d">https://medium.com/media/bb3dc31fe4641c24aefcab1dd47d3b1d</a>
‘/pricing/all’ endpoint will fetch the list of pricing details from the public schema and return the list.
‘/user/all’ endpoint requires a tenant identifier as a header and will fetch the data from that schema and will return the list of users.
Let’s insert some data into the tables
<img src="https://media.coffeebeans.io/blogs/0_Z-g0_JpRvYd0I3Cg.png" alt="dashboard 1">
<img src="https://media.coffeebeans.io/blogs/0_GKVCl0oNHp8gtUe2.png" alt="dashboard 2">
<img src="https://media.coffeebeans.io/blogs/0_Ve32K5o0OctpcC7c.png" alt="dashboard 3">
aapl schema
Let’s hit the APIs and see the results
We can use postman to check the APIs
<img src="https://media.coffeebeans.io/blogs/0_W-6OJ-AF2p8lMGbD.png" alt="postman pricing api">
list of pricing details from public schema
<img src="https://media.coffeebeans.io/blogs/0_Ym46lKGPjEKP2tD-.png" alt="postman users api">
list of users from google schema
<img src="https://media.coffeebeans.io/blogs/0_FTUlhar0yuykFvYy.png" alt="postman users api">
ist of users from aapl schema
Great! we finally achieved schema-based multi-tenancy for our application.
<h1>Important</h1>
For the simplicity of the implementation, we passed the tenant identifier in the header. It should not be passed in the header. A better approach will be to include it in the JWT tokens which can’t be modified and retrieve that by intercepting the HTTP request and storing the identifier in a TenantContext.
Now I hope you will apply this architecture to your SaaS product and save on your infrastructure costs.
<h1>Source code</h1>
You can find the source code <a href="https://github.com/vishalbihani1306/schema-multi-tenancy">here</a>.
<h1>Pros</h1>
<ol>
<li>Efficient use of resources.</li>
<li>A better architecture for multi-tenancy compared to the row per tenant approach in terms of data isolation.</li>
<li>Low infrastructure costs.</li>
<li>Can easily scale.</li>
</ol>
<h1>Cons</h1>
<ol>
<li>Development complexity is high.</li>
<li>High customization per tenant will be difficult.</li>
<li>Data backup for each tenant is not possible.</li>
<li>Monitoring for each tenant may not be supported.</li>
<li>Frameworks supporting this pattern may not be available in all languages.</li>
</ol>
Thanks for reading.

Architecture design is a must while developing a SaaS application to ensure its scalability and optimising infrastructure costs. In this blog, Lets discuss the implementation of one such architecture with Quarkus java framework and Hibernate ORM  By Vishal Bihani , CoffeeBeans Consulting

Expertise

Bits, Bytes, Blogs and more.