Rewriting Parse.com for scale
summary
Parse.com is a BaaS used by 300K apps and traffic grows by 10x per year so the team rewrites a Ruby monolith into Go and measures 5–7x faster in Go. Deploy times drop from ~35 minutes to minutes, traffic reaches ~15,000 rps, and a custom HTTP proxy reduces DB connections and exposes slow queries. Built features include shadowing live traffic to find real deltas, simple memcached throttles and auto-throttling, deciders (feature flags), fast parallel integration tests and in‑memory mongod instances.
AI-written summary · about
chapters (12)
- 0:00Introduction and context
- 2:41Identifying scaling issues
- 6:17Rewriting the stack in Go
- 9:44Monolith philosophy
- 12:26Implementing database proxies
- 14:30Shadowing live traffic
- 17:33Throttling and feature flags
- 20:52Deployment and tooling
- 24:50Context for request tracing
- 26:34Database management and internals
- 28:27Testing infrastructure
- 32:10Q&A and closing thoughts
related talks
47mScaling: from 0 to 25 million users
44mIntroduction to application scaling
54mGetting to 300 million requests a month with a one man AI/Dev-Ops team
40mBeyond your daily coding
37mHow Small Changes Can Make Huge Waves
54mPerformance Architecture
01h 03mMoving Fast - without crashing
25mFrom PHP and on-prem servers to NodeJS/React and AWS