Deploying hybrid topologies with Kubernetes and Envoy: A look at service discovery

Jose Nino (Lyft), Lita Cho (Lyft)

15:50–16:30 Wednesday, 6 November 2019

Location: Expo Hall Sessions

Expo Plus Sessions, Kubernetes, Overcoming Obstacles: Lessons in Resilience

Tags:

Who is this presentation for?

Software engineers and SREs

Level

Intermediate

Description

Before Kubernetes, Lyft had hundreds services that deployed as virtual machines (VMs) in AWS. And moving over to Kubernetes is a big undertaking.

Jose Nino and Lita Cho outline how Lyft leveraged Envoy and built a dynamic control plane (the service that hosts Envoy configuration and discovery API) to shape traffic between its legacy infrastructure and Kubernetes infrastructure. They demonstrate how Lyft handled service discovery within its hybrid environment, ways to control the traffic when Kubernetes issues came up, and how Lyft used Envoy as a sidecar container to keep in par with the legacy behavior. And you’ll leave with a view into the various Lyft faced with this migration plan, why it was needed due to the size of Lyft’s legacy infrastructure, and how those challenges were mitigated.

Prerequisite knowledge

Familiarity with Kubernetes and service-oriented architectures

What you'll learn

Learn how Lyft bridged legacy and new systems using Envoy

Jose Nino

Lyft

Jose Nino is the lead for core server networking group at Lyft. Jose has been instrumental in creating systems to scale Lyft’s Envoy production environment for increasingly large deployments and engineering orgs. He’s worked as an open source Envoy maintainer and has nurtured Envoy’s growing community. Recently, Jose has moved on to scaling Lyft’s network load tolerance systems. Jose has spoken about Envoy and other related topics at several venues, most recently at KubeCon EU 2018 and at the main stage at KubeCon NA 2018.

Lita Cho

Lyft

Lita Cho is a senior software engineer on the networking team at Lyft, where she builds out the service mesh to handle both Kubernetes and legacy systems. She also maintains the tracing infrastructure at Lyft. While at Lyft, she’s worked on building out the API infrastructure using protocol buffers, creating systems that generate code and bring type safety to Lyft’s polyglot microservice architecture. Previously, she was at DreamWorks Animation for four years, working on production and pipeline tools for artists and animated movies.