Phoenix 1.2 and Beyond
summary
Phoenix 1.2 adds a distributed presence built on CRDTs and Phoenix Pub sub, proving 2 million channel clients on one server and handling net‑split recovery without a central Redis. Presence uses an observe‑remove set with vector clocks to avoid tombstone blowup, ships server and client helpers for diffs and metadata, and exposes an accidental service discovery via presence for sharded, metadata‑aware process routing that works at runtime.
AI-written summary · about
chapters (24)
- 0:00Intro
- 1:55Growth of Phoenix
- 3:35Phoenix 12 release
- 8:04Redis vs Elixir
- 9:09Bestcase scenario
- 11:07CRDT Paper
- 13:22Selfdoubt
- 14:38CRTTs
- 16:55Time Illusion
- 17:28Alexander Sanj
- 18:30Vector Clocks
- 23:27Presence Tracking
- 24:27Presence Object
- 26:25Demo
- 31:09Replicas
- 34:02Beings Presents
- 35:14Accidentally
- 35:47Sending Sets
- 36:26Process Groups
- 38:04Client Load Balancing
- 38:54Game Sessions
- 39:30Service Call
- 39:50Custom Load Balancing
- 44:46Dockyard
related talks
33mWhat's Next for Phoenix
45mPhoenix a web framework for the new web
01h 00mPhoenix a Web Framework for the New Web
01h 00mPhoenix: an Intro to Elixir's Web Framework
23mCracking the PubSub Code. Elixir's Key to Scalable Real-Time Systems
41mElixir & Phoenix – fast, concurrent and explicit
32mWhy I choose Phoenix
29mRails to Phoenix