What news from AWS re:Invent last week will have the most impact on you?
Amazon Q, an AI chatbot for explaining how AWS works.
Super-fast S3 Express storage.
New Graviton 4 processor instances.
Emily Freeman leaving AWS.
I don't use AWS, so none of this will affect me.
Cloud Native Ecosystem / Software Development

Primer: The Who, What and Why of Service Mesh

A service mesh is a way to increase the observability, resilience and security in a large-scale containerized application. “Part of the attraction for working with a service mesh is that you get all three in one,” explains Varun Talwar, CEO of Tetrate and one of the creators of Istio.
May 22nd, 2019 3:00am by
Featued image for: Primer: The Who, What and Why of Service Mesh

If you’ve been hearing about service mesh — perhaps you’ve noticed that Istio seems to be everywhere — you might be wondering if you need a service mesh. Here’s a message everyone in-the-know agrees on, from the creator of Envoy, to the leaders of Tetrate and Aspen Mesh who are building companies on service mesh support to technology analysts: If you’re wondering about service mesh, you don’t need one. Period. If you’ve reached the scale and microservice maturity level that requires a service mesh, you will be actively — perhaps desperately — searching for a solution and it will be abundantly obvious that a service mesh is necessary. 

“I think we have a tendency to chase the shiny object, in the sense that X company does Y, therefore I must do Y, even though I don’t have any of X company’s problems,” explains Matt Klein, the creator of Envoy, an open service mesh.  

Service meshes solve legitimate problems — we’ll get into that in just a minute. Most companies just don’t encounter those problems until they are both working on a mature microservices architecture and are large enough to employ at least several dozen developers. “Companies like that are actually sort of the exception to the rule,” explains technology analyst Tom Petrocelli. “Your average bank doesn’t have massive clusters out there.”

But while most companies don’t need a service mesh right now, Petrocelli thinks it’s a market poised to balloon, because he sees the number of large-scale companies with mature, containerized applications increasing dramatically in the next several years. 

So what is a service mesh? What problems does it solve?

Observability, Resilience and Security

A service mesh is a way to increase the observability, resilience and security in a large-scale containerized application. “Part of the attraction for working with a service mesh is that you get all three in one,” explains Varun Talwar, CEO of Tetrate and one of the creators of Istio.

“There are a lot of common functions that you need to implement in every service, like monitoring, logging events, tracing requests, executing and encrypting all outbound calls, routing traffic to the service,” Talwar explains. In a service mesh, a sidecar proxy is attached to each service and handles all of these standard functions that every service needs. It’s called a “mesh” because the sidecar proxies handle all communications between the services in the application, creating a network “mesh” of services.

With the sidecar proxy handling all mundane tasks, developers no longer have to code those functions into the service. That frees up developers time to work on other things. More importantly, a large organization no longer has to rely on individual developers to ensure that communication is properly encrypted, that security certificates are rotated or that logging is properly set up. The observability the control plane provides also makes it possible to debug communication problems, as well as to get the information needed to pass compliance audits. “The application developer wants to focus on business logic, what makes the business money,” explains Andrew Jenkins, chief technology officer at Aspen Mesh. “They don’t want to implement the certificate rotation story over and over and over again. A service mesh means you only have to do it once and it’s part of your infrastructure.”

The service mesh is made of up two components: The data plane and the control plane. The data plane actually implements the service mesh and does the actual traffic routing, while the control plane has the user interface — it’s where you set the policies for the data plane and the lens through which you get the observability.

In some cases, the data plane and control plane components of a service mesh are decoupled, as is the case with Envoy (a data plane project that is hosted by the CNCF) and Istio (a control plane project created by Google that uses Envoy, but is not hosted by the CNCF). “It’s not true that you can mix and match anything, but there is some mixing and matching going on,” Jenkins explains. Linkerd, the other major open-source service mesh option, includes both the data plane and control plane together. Other players in the service mesh ecosystem include Nginx, and HashiCorp’s Consul.

The other way companies handle creating a standardized way to handle communications and observability is through client libraries. For smaller applications that only use one or two languages, this can be an easier way to ensure all the security, logging and communications code is always included, in a consistent way, in each service. In a real-world enterprise scenario, however, in which services are written in many languages, this technique tends to break down. 

The Future of Service Mesh

“I think eventually you will see service meshes in the majority of applications,” explains Petrocelli, the technology analyst. “I can’t imagine a bank implementing a cluster-based architecture and not having a service mesh. That would be crazy. Not only for the routing, but simply for the encryption and observability in the network layer.” 

As more big businesses move to mature, containerized systems, they will be looking for a service mesh to keep those applications secure and compliant. In the meantime, it’s a good idea to keep in mind that service mesh technology is still fairly immature, too. While Envoy is mostly production-ready, it’s still quite difficult to get up and running correctly. Istio, on the other hand, isn’t production-ready, in spite of Google’s massive promotionally effort — a fact both Petrocelli and Talwar, Istio’s creator, agree on. 

As service meshes become more common, they’ll have find a way to address issues like how to install upgrades, how to manage multiple meshes over multiple deployments and how to address the learning curve issues, including the fact that successful implementation requires more collaboration between developers and network engineers than you generally see at most companies. 

“There’s a lot of work to be done in service mesh, it’s early days,” Talwar explains. “Most people are talking value and are clearly sold on the value, which is good, but there are still challenges that need to be addressed.” 

Aspen Mesh is a sponsor of The New Stack.

Feature Image by Free-Photos from Pixabay.

Group Created with Sketch.
TNS owner Insight Partners is an investor in: The New Stack.
THE NEW STACK UPDATE A newsletter digest of the week’s most important stories & analyses.