Portworx sponsored The New Stack’s coverage of KubeCon + CloudNativeCon North America 2019.
One of the earliest entrants into the service mesh space with its Linkerd open source software, Buoyant has turned its attention toward the problems presented by building applications on top of Kubernetes that aren’t covered by a service mesh — namely, organizational problems. To that end, Buoyant has launched a private beta of Dive, which it describes as a SaaS “team control plane” that “extends Linkerd’s ability to solve real problems from the domain of services in an application to people in an organization.”
The company unveiled the new service at KubeCon + CloudNativeCon North America 2019, being held this week in San Diego.
“It’s directly related to everything we’ve done with Linkerd. The goal for both of these projects is kind of the same, which is, if you need to build a software application on top of Kubernetes, there’s a whole bunch of challenges that you have. Some of those challenges are very operationally focused and are best solved by something like a service mesh, but a lot of the problems are actually more problems of people and of process,” said CEO William Morgan in an interview with The New Stack. “That’s what Dive solves. We see Dive as extending, these same goals out of the realm of pure machinery and into the realm of who are the people at play and what do they have to know in order to coordinate and what tools do they need.”
Dive, said Morgan, has been a work in progress for the last four years, and gives its users a picture of their application “on a human level” — who owns what service, what service is a dependency for another, how all your services are related — and provides a sort of “Facebook for microservices,” he said. With Dive, each microservice has its own linkable page, as do all changes to infrastructure, that displays all of this information and can be commented on and shared with others in Slack, for example.
Beyond that, each page has information on where that particular code is hosted, when and by whom it was last deployed, and even service level objective (SLO) tracking. In fact, this aspect is a focus of Buoyant’s blog post, with the company writing that “just as Linkerd gives your infrastructure mission-critical reliability, security, and observability features, Dive gives your team the SRE superpowers they need to safely operate applications at scale.”
In its current nascent form, Dive requires a Kubernetes deployment running Linkerd, and integrates with GitHub, although Morgan said that a GitLab integration is on the shortlist. Written in Go and React, Dive gathers information from Linkerd to provide a number of features, including a service catalog showing where services are deployed, a record of every rollout across every cluster, a global service topology, automatic SLO tracking, service “report cards,” and deploy policies for services that consider SLO status and dependencies.
When we asked Morgan what Dive was built in, he originally said it was built in “Kubernetes” and “runs Linkerd.” He went on to explain that Dive falls into a class of application that, while it helps people running those technologies and dealing with their resultant complexities, it is also made possible by those same technologies.
“Dive is a natural extension of the service mesh. The value of the service mesh in my mind — and the same thing can be said of Kubernetes and containers — is less about what they do, and it’s more about what they unlock,” said Morgan. “They’re plumbing and what does plumbing enable? What kinds of things does it allow for? I think that’s ultimately the fun part of Dive, that it is the class of application that is suddenly enabled by the fact that we have all this cool open source technology under the hood.”
To sign up for the private beta of Dive, you can join the waitlist.
Feature image via Pixabay.