[SREcon Europe] PagerDuty's Service Mesh Migration

Talk Abstract:

Service Meshes are extremely trendy right now, and the benefits are many. Traffic management, flow control, multi-cluster load balancing, mTLS are all advertised benefits. But of course all these are benefits of having a service, and unless your company is brand new, nobody starts with one.

PagerDuty featured around 500 individual services needed to be migrated. What's worse, we had a very lean team to do it with. We'd use the "away team" model, where we made all the initial code changes required, then asked other teams nicely to review those changes.

It seemed like a nearly impossible task, but a year later we're looking at 95% completion and making good progress on the long tail. Along the way we've learned a lot about how to make a migration successful, what's good in a service mesh, and how to keep a team from burning out on an extremely monotonous task.

About the Presenter:

Liz Frost is a Senior SRE on PagerDuty's Core Infrastructure team. She's previously worked at Heroku, Heptio and Buzzfeed, and she's passionate about building platforms in any form they take. She's based in beautiful Vancouver, Canada.