Q: In 2016, KubeCon was all about Kubernetes, but by 2017, the community was really excited about service mesh (Istio, Linkerd, etc.). Why has it taken service mesh longer than Kubernetes to really gain mainstream adoption?
A: It was a combination of a few different things:
- Kubernetes gained some initial, early traction with users, but it took several years before many users had experience with it in production. Service mesh was always going to be the next step after Kubernetes gained traction.
- Kubernetes quickly became the defacto container management standard. Service mesh users had many options, and many of them tried different options before choosing one to move forward with.
- Several of the service mesh projects went through some architecture changes over the years, so maturity took a little longer. The good news is that service mesh adoption is now growing rapidly and many users are expanding their deployments.
Q: How has Istio evolved since those early 2017 days?
A: Istio has evolved in many important ways. First, it has the largest set of open source community contributors, so it’s benefitted from really great engineering and ideas. Second, many companies are using Istio in production, across all the private cloud and public cloud Kubernetes platforms. Third, it has a long history of continuous innovation and new functionality. Fourth, it’s really setting the new standard for flexibility with the announcement of Istio Ambient Mesh. And finally, having Istio be accepted as an official project of the CNCF means that there is no longer any question about its future, and we expect a whole new set of developers to start contributing interesting ideas to the project.
Q: What does Istio Ambient Mesh mean for the Istio community and Istio users?
A: The Istio community has always been really good at listening to users and adapting the technology to their needs. One of the areas that the community heard from many users was that the next step in their Kubernetes journey was service mesh, but an interim step in that evolution was getting zero-trust security in place. Istio Ambient Mesh will make it much easier to take that initial step with service mesh, and allow Kubernetes users to continue to make progress in their evolution with microservices. It’ll also set them up to take advantage of all the other powerful things that Istio enables – security, observability, business logic offloading for application teams.
Q: Should users be paying attention to the sidecar vs. sidecar-less discussions that are happening across various service mesh communities these days?
A: Yes and no. The reality is that sidecar or sidecar-less is an implementation detail, and may or may not be relevant to certain teams or certain applications. These options give users choice around a few important elements – cost, operational complexity, and performance. And different applications need different things in layer 4 and layer 7 from a service mesh. I think the nice thing about what the Istio community has done is to give users choice on a granular level. They don’t force a service mesh architecture on the users, but rather they allow the users to pick the right configurations for their applications and their operations teams.
Q: What are you excited about in the service mesh space as KubeCon + CloudNativeCon in Detroit approaches?
A: First and foremost, we’re really excited to see the community in person again. KubeCon Europe in Valencia was great, but KubeCon North America is always a bigger show and it’s been several years since we’ve gotten together in person. So many great ideas happen when we’re all together. Next, we’re excited to hear more feedback about Istio Ambient Mesh. The initial feedback has been exciting, so we’re eager to hear from a broad set of users. Third, we’re curious to see the evolution of Gateway API and how it interacts with both Kubernetes and Istio. We’ve seen many users start to bring together how they think about API Gateway / ingress and how they think about service mesh. And finally, eBPF is a very exciting area that has potential impact on service mesh, security, and observability. It’s still an evolving technology, but it brings interesting potential to cloud-native applications.