Architecting for scale
summary
Argues for starting simple and proving correctness before adding scale infrastructure, with concrete outcomes: a legacy system that stored everything in the database and accumulated half a million stored procedures becomes untestable; Prime Video’s monitoring prototype moves from many Lambdas to in‑process sharding and achieves a 90% reduction in cost; Stack Overflow runs a simple single‑deployment stack serving 30M searches/day while Shopify reaches ~1M RPS with a monolith; advice: stay in process and use telemetry to scale just‑in‑time.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 3:11Why scale and simplicity
- 5:42Not about
- 8:11Hidden costs of scalable designs
- 12:28Problem with design interviews
- 16:20Real-world scaling approaches
- 21:18Key principles for scalability
- 24:26Takeaways and final thoughts
related talks
49mLarge-Scale Architecture: The Unreasonable Effectiveness of Simplicity
46mDigital scale-up architectural patterns
48mMinimum Viable Architecture
14mGrowing pains: Scaling and re-architecting systems under fire
29mApps, Stacks, and Frameworks: Avoiding “Shiny Object” Syndrome
18mAPIs at Scale Through Minimalism in Code and Architecture
56mScaling Up!
49mSoftware Architecture, Team Topologies & Complexity Science