Messaging @Scale
summary
Platform balances agility versus efficiency by turning messaging into an ecosystem rather than a simple service: Msys uses SQLite-backed relational tables (chats, messages, presence) and a custom DSL CQL that compiles to C with codegen to Java/Objective‑C. A 2020 messenger rewrite makes the client 1/4 the size and twice as fast, Msys integration into Facebook cuts time‑to‑interact by 25+%, and the platform lets apps pick tables and stored procedures to avoid one‑size‑fits‑all bloat.
AI-written summary · about
chapters (11)
- 0:00Mobile messaging platform intro
- 1:09Messaging features in Meta apps
- 2:54Agility versus efficiency
- 3:37Divergent vs converged codebases
- 6:43Agility and efficiency balance
- 7:04Platform model: service vs ecosystem
- 9:13MSUS: messaging system platform
- 11:21MSUS integration in Facebook
- 12:46Thread view reuse
- 13:30Scaling MSUS challenges
- 14:55MSUS future enhancements & Q&A
related talks
17mScaling Messenger Product Development
53mMessaging: The fine line between awesome and awful
30mRebuilding Facebook Chat
01h 01mMessaging: The fine line between awesome and awful
36mTammy: a modular Matrix messenger
24mBuilding and maintaining messenger.com
46mFrom a Ruby on Rails Monolith to Elixir and Elm Microservices
31mSimplified Sliding Sync