SRE Weekly - SRE Weekly Issue #435

CrowdStrike released a lot more discussion about what happened widetailth their bad deployment, and yet there's still a frustrating lack of detail on the actual cause of the blue screens.

  CrowdStrike

A story of how properly positioned rationales can be powerful enough to prevent prod incidents

And a great place to put that rationale is in your git commit, says this article.

  Jean-Mark Wright

Need to choose between Redis and Memcached? This one's for you, with a qualitative comparison and relative performance numbers.

  Rahul Chandel

How do you promote interactions between fans without exploding your system with n-squared worth of messages, where n is the number of users?

  Matthew O'Riordan — Ably

If you want to convert to serverless, don't switch to microservices or change your datastore at the same time, argues this article.

  Yan Cui

It's all about the owl's butt (and sending 4 million push notifications in 5 seconds).

  Zhen Zhou — InfoQ

Here's the second part of the article I wrote on my recent project at work, taking down a full AZ in production.

  Lex Neva — Honeycomb

  Full disclosure: Honeycomb is my employer.