How the Service Mesh became a Service Mess

[weaveworks]

The nature of technology is to continuously evolve. Each iteration draws on learnings and mistakes of the past. As we move past the cusp of change from monolithic infrastructures to cloud based microservices, we are experiencing an evolution of the tooling and monitoring of our applications. We knew how to monitor our servers, network devices and our applications running on our systems, however the shift to microservices requires a new infrastructure and with that new ways to automate and monitor it.

In the early days at Twitter, William Morgan, worked heavily on their in-house tools to monitor their transformation to a microservices based infrastructure. One critical tool, called Finagle, was eventually open sourced and became the basis for Linkerd which William and co-founder Oliver Gould eventually dubbed a “Service Mesh”.

...we were looking at Finagle and we were saying this thing was so transformational at Twitter that we want to bring this to the rest of the world…

Linkerd is reminiscent of a monitoring system developed at CloudFoundry leveraging some of Netflix OSS releases as well as the Spring Framework. Without out of the box tools, innovators built what they needed without knowing it was a service mesh.

Now with the widespread adoption of service meshes to manage microservices infrastructures, Buoyant has watched Linkerd adoption skyrocket among organizations adopting cloud native technology such as Kubernetes. (Here is an excellent article by William explaining what every Software Engineer needs to know about service meshes.)

The value of the service mesh...by moving that stuff out of the libraries and into proxies that can be attached at run time...you actually empower the platform team to own those features, so you're now providing the features at the platform level in a way that the application and developer teams doesn’t have to be aware of. So that decoupling of platform and developer teams is actually the real value of the service mesh.

Tune in for a brilliant discussion on the origins of service mesh, its ecosystem and why it’s important for Kubernetes centric infrastructures today.

William Morgan, CEO Buoyant

William Morgan is the CEO of Buoyant and one of the creators of Linkerd. Prior to founding Buoyant, he was an infrastructure engineer at Twitter, where he ran several teams building on product-facing backend infrastructure. He has worked at Powerset, Microsoft, adap.tv, and MITRE Corp, and has been contributing to open source for over 20 years.

To be notified of future episodes:

Listen to the full episode:

Apple Podcasts | Spotify | Stitcher | SoundCloud | Pocket Cast | Google Podcast | Overcast

The Art of Modern Ops · How the Service Mesh became a Service Mess

View the full article