High availability with Elixir and Erlang
summary
Claims that Erlang/Elixir give concrete runtime primitives for real high availability: lightweight processes use about 2KB per process so systems run millions of processes concurrently; message passing with no shared memory and 1ms scheduler quanta prevents single tasks from stalling the whole system; supervisors form supervision trees that automatically restart workers; observer tool exposes live state over SSH; built systems achieve hot code upgrades and self‑healing restarts, not just theoretical patterns.
AI-written summary · about
chapters (10)
- 0:00Start
- 0:22High Availability with Erlang
- 10:30Request Response Patterns
- 10:54Server Processes
- 15:51Run Time
- 19:39Supervision Trees
- 21:56Observer
- 28:34Resources
- 29:13How Do You Think Elixir / Erlang Relate to Http-Based Micro-Services Architecture
- 31:27What Is Elixir Not Good for
related talks
59mKeynote: Elixir Should Take Over the World
41mWhat NOT to do when building large scale systems in Erlang
01h 00mKeynote: Over a Century of Programming
57mMaximising Side Effects When Adopting Functional Programming
51mKeynote: If at first you don't succeed (..)
47mThe art of challenging assumptions
39mKeeping Code Consistent
40mKeynote: The Pendulum