A Brand new Immune System for a Brand New Google Product (Hebrew)
The full story behind a highly successful immune system we implemented for the "New Google Sites".
We will discuss key decisions and present the design of the system that we built, touching a wide range of devops topics:
Production tests
Webdriver vs. RPC probers
What environments we have and how are they pushed (CI/CD, etc.)
Capacity testing, load testing, Canarying
What worked, what went wrong, and what were the surprising wins
Blackbox vs. whitebox alerting
Capacity planning and request costs
What the oncaller dashboard looks like
Postmortem culture