Service Mesh Is Still Hard - The Things We All Did Wrong & Tales of Woe

Explore the challenges and complexities of service mesh implementation in this 30-minute conference talk from KubeCon + CloudNativeCon. Delve into the tradeoffs, common mistakes, and lessons learned in the rapidly evolving service mesh ecosystem. Gain insights on successfully building and running cloud-native applications, with a focus on Linkerd examples. Learn about meshing pods, routing traffic, implementing mutual TLS, and monitoring mesh health. Understand the importance of maintaining boundaries, handling TCP and hosts, and addressing bugs. Discover when a service mesh is necessary and how to analyze its effectiveness. Examine the possibilities of zero configuration and the realities of code changes. Acquire knowledge on assessing both application and mesh performance in this comprehensive overview of service mesh complexities.

Intro
Speaker Introduction
The service mesh tradeoff
This is a talk about complexity
Who pays the complexity cost?
Lots of Linkerd examples
Example: meshing a pod
Example: routing traffic
Example: mutual TLS
Example: mesh health
Concepts are not obvious
Maintain Boundary
TCP and Hosts
When You Find A Bug
Health checks
Services uses https
Do you need service mesh?
Need Mesh Analyzer
Zero configuration is possible
Zero code change is not!
How is my application doing?
How is my mesh doing?
Conclusion