Memphis.Dev case study
5 min read
Many organizations are moving their applications to a microservice-based architecture as it provides a lot of benefits over the traditional monolithic architecture. One benefit of this microservice-based architecture is greater response speed.
However, to achieve this, we need to first make sure that our microservices are communicating with each other efficiently. First, let us look into how these microservices communicate.
How do microservices communicate
Microservice-based architecture means that your application is broken down into a lot of small components and each component handles one small task of the entire application. Let's take a text messenger for example. One microservice will handle how the messages are sent, another one will handle receiving the message, another one will store the messages in a database, and so on.
In the above example, we used a very simple application that can send, receive, and store text messages. Now imagine that you have a more complex application that handles audio messages, videos, images, voice calling, video calling, and many other features. This would introduce a lot more microservices into our application.
If we want our application to work, which we obviously want, we need hundreds of microservices to talk with each other, and we want them to do it efficiently. There are two main ways that these microservices would communicate with each other. Synchronously, or asynchronously.
Syncronous vs Asynchronous
If you've worked with any aspect of web development, you already know the difference between Synchronous and Asynchronous communication. We'll quickly go over the two.
Imagine that we have a network for five microservices that need to talk to each other for different tasks. Let's say service 1 needs to talk with service 5, service 2 needs to talk with service 3, service 4 needs to talk with service 5, and service 1 needs to talk to service 2, all at the same time.
Notice that service 5 is talking with both service 1 and 4 at the same time. If we take a synchronous approach, then service 5 will first respond to service 1, since the request was made first. Meanwhile, service 4 will not get a response, and service 5 is talking with service 1.
You can see the downside of this when we have nearly a hundred services. This will impact performance a lot.
On the other hand, with async communication, service 5 will not wait until the communication with service 1 is complete. It will execute both requests at the same time.
This does help the performance of the application, but this method is quite inefficient because many requests are being made at the same time, and if the number of requests exceeds the number of requests that the service can handle, the service will crash and our applications would experience some downtime. We obviously do not want this to happen.
What is a message broker?
To solve the problem we mentioned above, we would use something called a message broker. Let's look into what exactly is a message broker, and how it helps to solve the problem.
Let's use the example of the five microservices that we used previously. We are still using Async communication but this time, instead of sending the requests directly to the services, and the services listening for requests from multiple sources, we direct everything to a single source. This single source is what we call the message broker.
Every single API in our application will send requests to and listen to this one central message broker. To understand this let's take an example.
Let's say that service 1 wants something to be done. So it will send the request to the message broker. Every other service in the network will be listening to this message broker, which will inform the services that "Service 1 has made a request to do something." Now let's say that service 3 knows how to do what service 1 is asking. Hence service 3 will go ahead and do what service 1 is asking and will send a message back to the message broker saying "Hey message broker, I've done what service 1."
Remember that this is still an async communication. So service 1 does not care whether the work is done or not.
This entire system of handling async communication using a message broker makes the entire process more efficient, and it also becomes easier to troubleshoot any issues.
Memphis.dev
Memphis. dev is the only low-code real-time data processing platform that provides a full ecosystem for in-app streaming use cases using Memphis distributed message broker with a paradigm of produce-consume that supports modern in-app streaming pipelines and async communication by removing frictions of management, cost, resources, language barriers, and time for data-oriented developers and data engineers, unlike other message brokers and queues that requires a great amount of code, optimizations, adjustments, and mainly time.
What are the unique points
If you've worked with data streaming before, you know there isn't one solution for all your problems. You would need to use multiple solutions such as Apache Kafka, Flink, and NiFi together to get the desired result. This helps you achieve almost real-time data streaming.
With Memphis, you get proper real-time data streaming which in turn helps in improving application performance. It is also much easier to install and start using than combining various solutions together.
Memphis also has a focus on resiliency meaning you don't need to worry about losing data because of a lack of retransmits, crashes, and monitoring. It also solves all the challenges we mentioned above with async communication and helps you to debug your application easily.
Conclusion
Message brokers can help in achieving greater performance with your applications, but to implement them properly, you need to combine many solutions together.
Memphis handles all the tasks and provides greater performance and resiliency compared to stitching multiple applications together and overall provides a better developer experience.