• No More Blue Fridays: Make deployments safer my using different techniques including canary testing in production, staged rollouts and resilience engineering
  • TigerStyle! (Or How To Design Safer Systems in Less Time) by Joran Dirk Greef: Talks about several things you can do to make a system safer. 2 that stood out to me are adding limits / constraints and the use of assertions. (If something terrible is happening consider crashing the system … :))
  • Postcards from the Peak of Complexity by Brian Goetz: Brian Goetz talks about what goes into designing a major new java feature. Doing hard things follows goes through phases similar to stages of grief. Halfway through you reach the “peak of complexity” but this often isn’t the end state. You often have to try some number of bad ways of doing it before you can know the right way.
  • Distributed DDL Replication at Global Scale by Gwen Shapira: A company providing postgres as a service describes how they execute create table type commands across many db instances. They run in a multi-tenant with multiple replication groups since not all customers can fit in a single db cluster. Lots of energy spend on failure testing. (Data corruption is probably bad for a db company)
    • “Not a fan of chaos engineering. If you hit an obscure error you can’t reproduce you just have noise. Prefer deterministic testing.” - This is an interesting take!

I will go back and watch srecon2024 americas videos. Lost track of those at some point but it’s good to keep this stuff front of mind and to have it repeated. :)