Microservices at Netflix Scale: Principles, Tradeoffs & Lessons Learned - R. Meshenberg

https://www.youtube.com/watch?v=57UK46qfBLY

Principles

Principles in action

stateless services

register to service discopvery

health check

get external stuff from service discovery

-> test it in prod

billing services (needs to be just right)

limited access

auditable

===

COST

Need for microservices

1) Contract for intermediate language (IPC)

2) protect your dbs

3) operational visibility

is it scalable? get signal out of noise

Failure, need to test in prod

chaos monkey

cf kanban "crash and burn"

containers != microservices

containers at scale

- spinnaker

CI/CD

===

SUMMARY

value

org change

your own context

===

QUESTIONS

If chaos kong kills a region, and nobody notices, did it happen?

telemetry knows

engineers need to check telemetry regularly

How to decide which services new/remove

team itself

Destructive testing on financial parts.

failure will happen

is tech ok?

are ppl ok?

Service visualisation, where can I get it

flux and ...

What discipline do you require from engineers

freedom & responsibility

Cross service functionality

eg. security