Beyond the Basics: Production Serverless Patterns for Extreme Scale
summary
Explains how AWS Lambda reaches 1000 concurrent environments immediately and adds roughly 1000 more every 10 seconds, reaching ~16–18K execution environments in a few minutes (translating to tens of thousands RPS), how provisioned concurrency eliminates cold starts with scheduled and target auto-scaling and a 16% duration discount on good utilization, how Graviton yields ~20% lower price for ~34% better price/perf, recommends Power Tuning and AWS “countdown,” and notes provisioned concurrency spills to on‑demand so there’s no availability drop.
AI-written summary · about
chapters (8)
- 0:00Intro
- 1:26Let's talk about scale
- 3:34How do we solve for bursty workloads?
- 5:59Let's dive deeper
- 12:13But what happens after your big event?
- 18:31But what happens during an unexpected traffic spike?
- 19:39Summary
- 20:28Outro
related talks
51mArchitecting for Hyperscale with AWS: Navigating Real-world Serverless Challenges
19m(N+1) Lessons I learned about serverless
44mServerless Architecture on Production: Good, Bad and Ugly Parts
50mServerless Architecture Patterns and Best Practices
29mBuild highly scalable serverless applications on AWS
40m4 Key takeaways from running Serverless on production for 4 years
17mCloud Computing Made Better: An AWS View on Serverless Computing
47mAdvanced Serverless Architectural Patterns on AWS