Building for Scale: How not to get Caught out at Global Launch
summary
Metaplay’s backend pattern focuses on preventing post‑launch chaos by planning for real scale: avoid pure stateless/serverless designs that push traffic into a persistence bottleneck and trigger enormous cloud DB bills; expect surprising peaks (example: cases approaching 100 million players). Swap JSON configs for prebuilt binary archives to prevent 30 second loading times and crashes on old Android devices. Build a versioned config pipeline from spreadsheets, enable progressive rollouts, A/B, rollbacks, and make components replaceable so scaling is an extension, not a rewrite.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:31Speed to market isn’t enough
- 3:15Four pillars of scaling
- 4:39Pitfalls of quick systems
- 6:25Expanding for global success
- 8:08Hope isn’t scaling
- 8:51Design for scale
- 10:37Game configs scalability
- 13:53Speed versus scalability
- 14:57Meta Playay’s tech solution
- 15:39Q&A
related talks
32mScaling Done Right
45mScaling under uncertainty
46mDigital scale-up architectural patterns
50mScaling for Global Growth with Modern Cloud
48mScaling for Global Growth with Modern Cloud
56mMastering Cloud Scalability and System Designs Deep dive
19mEngineering for the Long Game
26mCreating a scalable ecosystem of Microservices