What is Service Mesh and Why you should consider it

Brief about Software Architecture History

Before to start the explanations about Service Mesh infrastructure, let’s understand a little bit about what we have created before, as software architects and developers.

I think it is essential to know why Service Mesh might be useful for you.

We’ll talk a little bit about Software Architecture History. I promise it will be rapid, but it will be worth it.

Let’s analyze the image below

There are different architectures models in the Timeline. The MVC Pattern which one is present in our architecture today.

SOA and EDA were popularized in 2000 and 2003, respectively; things started to be distributed. We started at that time to break things into small pieces of software, for several reasons.

And finally Microservices and Serverless, in 2013 and 2015 respectively, and different approaches about software development were coming.

Conway’s law and Inverse Conway Maneuver sprung up to the scene to explain how to work with Software Architecture in terms of business and Teams.

There is typical behavior if we look at this Timeline, we are trying to divide our software pieces into small and smaller parts, as small as we can.

But, what is it important or related to Service Mesh Infrastructure???

Microservices are distributed systems, and distributed systems means handle network problems.

That is exactly what Service Mesh Infrastructure can help us. Abstracts network issues from developers.

Service Mesh

In a nutshell.

Service Mesh can be defined as a dedicated infrastructure to handle a high volume of traffic based in IPC (inter-process communication). In a microservices architecture, usually called East-West Traffic.

In a few words, service mesh can be considered a “layer” to abstract network for services communications.

These abstractions solve the most part of the network handlings like Load Balancing, Circuit Breakers, Retries, Timeout, Smart Routing which can enable advanced deployment techniques like Canary Releases, Dark Launches, and others.

Then, we can take off these responsibilities from our application code. Also, we can remove these roles from developers, and it is very important because developers should code for the business, not for infrastructure requirements.

Another important characteristic of Service Mesh is Telemetry, some implementations integrate with Jaeger and Prometheus easily.

Famous libraries in the Java ecosystem related to network handlings like Netflix Ribbon, Hystrix and Eureka can be replaced for the Service Mesh implementations like ISTIO.

Service Mesh & Microservices Architecture

In general, in the Microservices Architecture, service-to-service communication is quite complex.

Usually involves different communications patterns like REST and gRPC over HTTP or AMQP for asynchronous and durable communications.

As we can see, microservices are distributed systems, that is the reason why Service Mesh Infrastructure fits very well.

Practical example

Let’s look in a simple and pretty standard Microservices Architecture

Standard Microservices Architecture

There are some important things to look here.

North & South Traffic

North & South traffic usually happens between different networks look at the image Network A and Network B. This kind of traffic comes from outside our infrastructure, our clients, and this traffic is not trusted because the external network is out of our control.

We need heavy security here, that’s our Gateway to protect our applications.

Usually, we have an API Platform to manage our external APIs. API Management techniques and processes can help us with this task.

East & West Traffic

On the other hand, the East-West traffic happens in general on the same network, as we saw before, normally it is called service-to-service communication or IPC.

That is the place where Service Mesh lives.

gRPC is a very interesting framework if you are looking for high throughput applications or service-to-service communications.

Conclusions

Service Mesh is an interesting thing you are trying to play with Microservices Architecture, but I strongly recommended you to understand a little deeper before adding Service Mesh in your Architecture Stack.

There is no silver bullet when you think about Software Architecture but we as Software Architect, Developers and other need to understand and propose the right solution considering the company context.

Kubernetes Patterns – Sidecar

Motivation

On the last week, I’ve blogged about Ambassador Pattern.

This pattern is very important when we are trying to solve network issues in the Microservices architecture, in a few words Ambassador is a kind of proxy, to help in the service-to-service communications.

Today we’ll talk about Sidecar Pattern, it’s an interesting pattern when we are looking for help with network issues, but as we will see during this post, there are more features which this pattern enable for us.

 

Context

In the containers world, we need to follow the container Golden Rule, the container should have one single purpose to exist. That is the most important thing to follow.

When we are developing applications using the microservice as an Architectural guide, we shouldn’t worry about concerns related to infrastructures, like log collector, network handlings and other orthogonal concerns. These concerns are more related to the platform where we are running our service than our application code.

We should use our platform to help us with these activities. Kubernetes is a “de-facto” platform to run containers workloads. We can use Kubernetes to deploy a dedicated infrastructure to handle internal network traffic, ISTIO for an example. In this case, ISTIO is our “platform” to help with network handlings.

I’ve blogged about my first impressions about ISTIO and Service Mesh

Kubernetes has the primitive called PODs, the small unit of computational resources in the kubernetes ecosystem, the POD is able to have multiple containers, in that scenario the Sidecar Pattern is a perfect solution to help the main container.

Let’s look at the POD anatomy (the yellow one)

Solution

The Sidecar container should add some additional functionalities to the microservice container. The important part to pay attention here is the sidecar should run in a different process and is not able to change anything in the microservice container.

In the same POD, containers are able to share the volumes and the same network, it means the containers can reach each other via “localhost” for an example.

Let’s analyze an example.

In the real microservices architecture, we might have different services and many instances of these services, but, how we are able to look effectively at the logs?

We need a centralized tool that collects these logs, also we need an effective way to query these data to find something that helps us to troubleshoot and debug distributed systems.

Is that role of the main container to send these logs for service in the cloud? Maybe a Sidecar container is able to collect these logs, they are sharing the volumes, and send these data to the cloud.

The sidecar “enrich” the main container functionalities sending data to the cloud systems. That is the main role of Sidecar Container.

Look at the image below:

As we can see, the logger container sends the data to the cloud storage, the logger read data from POD volumes, because they are sharing the disk.

The microservice container doesn’t care about the logs, the main container should play to service our business only.

That is one example where sidecar container can help us adding extra functionalities for our main container.

In the Service Mesh Infrastructure, the sidecar container can help us adding some extra functionalities to help us to handle network issues, it is another example.

Conclusion

The Sidecar Pattern is very useful when we are working in distributed systems, especially in containers world.

It will increase our productivity because we don’t need to pay attention to infrastructure stuff and it makes our code more concise than ever without infrastructure handlings.

Then, it is time to say goodbye to Netflix Ribbon, Netflix Eureka and Netflix Hystrix and put their responsibilities to sidecar container.

References

Kubernetes Pattern Book

Microsoft Azure Docs

Kubernetes Patterns – Ambassador

 

Motivation

Recently, I’m studying kubernetes in-depth, mainly in part about how to use platform features to help me to work with distributed architectures.

During this journey, for my surprise, I’ve found many books of Kubernetes Patterns, and my god, these books opened my mind about “How to use Kubernetes effectively”.

My favorite one is Kubernetes Patterns, the book is awesome, it’s a kind of guide for me right now. The book describes many patterns and categorizes them in principles like Predictable Demands, Declarative Deployments, Health Probe, Managed Lifecycle and Automated Placement.

Today, I’ll talk about an important pattern related to network management techniques.

Let’s talk about Ambassador Pattern.

Ambassador

Context

When we are working with distributed systems, the network is the biggest challenge to solve, remember The Fallacies of Distributed Computing.

We need to do an effective strategy to work with outages, service discovery, circuit breakers, intelligent and dynamic routing rules, and time-outs.

In general, these things require a lot of configuration files envolving connection, authentication and authorization. These configurations should be dynamic as well, because in the distributed systems, addresses for instances, changes a lot during a certain timebox.

Of course, sometimes we are not able to handle these issues because our “application” is not able to handle it, our framework which the application is coded doesn’t support these features.

Also, we need to remember the containers Golden Rule, the container should exist for one single and small reason.

Maybe, handle these challenges into our application code cannot be a good idea, especially because sometimes we need to integrate with legacy applications.

The ambassador help us exactly at this point, let’s see how it happening.

Solution

Ambassador acts as “proxy” and hides all the complexities of accessing the external services.

We will put the ambassador container between our main application and external services connections. Just to remind, our ambassador container should be deployed in the same Kubernetes POD, which resides our main application container.

Using this simple approach we able to handle network failures, security, resiliency in the ambassador container, simple and effective way to handle these hards things to solve.

Look at the image below

The Ambassador Container should handle configurations related to Service Discovery, Time-outs, Circuit Breaker, Smart Routing and Security

Conclusion

The ambassador Pattern is very useful when we are working with distributed systems. It will reduce our main application complexity taking off the network management in our application code.

Remember: It will add some latency overhead. If network latency is a critical point for you, maybe you need to think about the ambassador adoption.

 

References

Kubernetes Patterns book

https://docs.microsoft.com/pt-br/azure/architecture/patterns/ambassador

Releases, Deployments and Traffic Mirroring

During my journey to learn ISTIO and your stack I’ve discovered some interesting concepts about deployments stuff. The first one I didn’t know the difference between Deployment and Release if you know no problem I’ll explain detailed during this blog post. Also on the next post, I’ll explain how ISTIO can help us to achieve it.

Deployment vs Release

The first thing to know, before deep dive in strategies is to understand the difference between these concepts. I’ve discovered it reading the excellent Christian E. Posta book Istio in Action. The book is under production then there is some chapters to release yet.

Deployment

Deployment can be described as an activity to install new code into production or another environment at runtime, the important thing here it can’t affect users anyway, or we can’t change traffic to these artifacts. Then we can deploy multiple versions without problems.

Release

A release can be when change traffic to a deployment did previously; it can affect system users, then we should plan it carefully. There are some ways to minimize the user’s impact during our releases. We’ll discuss it detailed in this blog post. Also, the version avoids “Big Bang” deployments, like blue-green deployments.

Request Level or Traffic Shifting

Now we know the main difference between Deployment and Release, then we can discuss another critical topic Requests Distributions Strategies or how is the best strategy to split traffics during deployments.

There is two ways to achieve it . Traffic Shifting or Request Level it is super important to understand because based on that you should choose the best option in your use case.

Request Level

It is a kind of self-explanatory, in this technique, we can split traffic based on request headers attributes and then control production traffic as we want. This strategy allows us to gained more fine control in production traffic during our deployments.

For example, we can change traffic based on client-id, in OAuth protocol, when this specific client can be a partner to test our application in the real world.

Traffic Shifting

Traffic shifting can be an excellent option when we did not expect to “identify” users by something in the request. In this strategy, you, want to split traffic to different based on a percentage of calls. This strategy is a little bit more simple that Request Level but can be an exciting option to test our deployments.

Let’s talk about Releases Strategies!!!

Dark Launch Releases

In this kind, we can change the traffic to a new deployment using the minor part of users based in some rule, a percentage for instance. The important note here is the most of users a.k.a production traffic should go to the “stable” version. The main idea here is testing new features to a set of premium users and then measure the adoption or something important for your company.

Let’s see an example using the Traffic Shifting Strategy

Dark Launch Example

Canary Releases

The idea is very similar to Dark Launches, but there is a small difference, in the Canary Release we want to test a new version of our deployment, see performance and system behavior. In this kind, it is not related only something new, feature or significant change. Sometimes we want to test new versions which one has performance improves for example. In this example we’ll use the Request Level Strategy, let’s see it.

Canary Release

On the example above, we change the production traffic to a new version only for client-id = 10. Others client-ids go to stable version of our application.

Traffic Mirroring

The idea of Traffic Mirroring is pretty simple. We will route the real production, a copy of the production requests to a new deployment or experimental version. The copy of the request is based on fire and forget principle and won’t impact the real user’s requests. Mirroring traffic is an interesting techniques to delivery code into production with more confidence.

The image below will show the Traffic Mirroring strategy

Traffic Mirroring Flow

These concepts are very important to know. It helps us to choose the correct strategy during our deployments.

In my opinion, this knowledge is the key point to guide us to choose an successfull deployment strategy.

On the next post we will learn how to do it using the ISTIO an open-source service-mesh implementation.

References

ISTIO in Action by Christian E. Posta

Blue Green Deployments by Martin Fowler

Install ISTIO on AZURE AKS

Hello,

During in my learning path to understanding Service Mesh and ISTIO. I decided to use some different cloud vendors. I choose Azure and Google.

I’ve started with Google Kubernetes Engine (GKE). It was my first experience with Google Cloud Platform components and was amazing. The command line is well documented and easy to interact with Kubernetes APIs.

Today I will explain how to install Istio components in Azure Cloud (Azure Kubernetes Service or AKS) which one offers managed kubernetes in Azure Infrastructure. On this post, I will use HELM to install Istio on kubernetes.

Let’s start with some requirements:

  • HELM Client ( installation instructions can be found here )
  • Azure CLI  (installation instructions can be found here )
  • kubectl ( installation instructions can be found here )

Creating the AKS Cluster and Preparing HELM

To create the AKS Cluster we can use the following statement:

Some considerations about this command:

  • I strongly recommend creating your own resource group
  • The –enable-rbac is mandatory to deploy ISTIO.

Then we need to configure our kubectl on Azure we can do it using the az command line, like this:

Now our kubectl is fully configured, we can start to install ISTIO in our AKS cluster.

Let’s start downloading the Istio Release. The zip can be found here. We are using the 0.8.0 version which one is the stable version. You need to choose the target OS, in the cluster.

Go to the ISTIO root folder and then we need to create a service account for HELM, it can be done using the following command:

Good, you should see the following output:

Awesome our service account is ready.

Let’s deploy our tiller deploy. Run the command above:

Then we can see, the following output:

Awesome, our HELM client is ready to start to deploy Istio.

Installing ISTIO

Go to the ISTIO root folder and then we can deploy install ISTIO in our Kubernetes cluster. It can be achieved with this command

After we can check the ISTIO components using the following command:

All the pods need to stay in Running state like in the image below:

Well done, our ISTIO is installed in our cluster and ready to receive some microservices.

On next week I will explain how to interact with our cluster, creating some microservices and manage the cluster monitoring tools like Grafana, Jaeger and other.

References:

Install ISTIO: https://istio.io/docs/setup/kubernetes/helm-install/

Create a cluster in AKS: https://docs.microsoft.com/en-us/azure/aks/kubernetes-walkthrough