Seeking SRE

Chapter 26. The Service Mesh: Wrangler of Your Microservices?

Matt Klein, Lyft

Over the past 5 to 10 years, microservices have become all the rage in distributed systems design and operation. Once relegated to only the infrastructures of the largest internet companies, the technosphere now buzzes with phrases such as “immutable container provisioning and scheduling,” “continuous integration and deployment,” “decentralized control,” and “polyglot language implementations.” It is absolutely true that microservice architectures do allow large development teams to operate with more agility. However, often overlooked is the harsh reality that large internet companies have typically invested hundreds of person-years in development and operational effort to make distributed architectures work well in practice. Reliability engineers are left wondering how to grapple with the myriad operational problems that come up when attempting to move away from our monolithic applications and deploy such an architecture in practice. How do services find and communicate with one another? How are distributed services observed and debugged? How are services packaged and deployed? What kind of complex failure scenarios can happen?

As on-the-ground microservice practitioners soon realize, the majority of operational problems that arise when moving to a distributed architecture are ultimately grounded in two areas: networking and observability. It is simply an orders-of-magnitude-larger problem to network ...