A GOTO 2025 talk by AWS Lambda's product lead covering how serverless handles extreme traffic spikes. Explains Lambda's concurrency model (cold starts, execution environment reuse, 1000 concurrent environments instantly plus 1000 more every 10 seconds), provisioned concurrency for steady-state workloads with scheduled and target-tracking auto scaling, and cost optimization tips including power tuning, Graviton instances, and disabling provisioned concurrency on test workloads. Key takeaway: serverless developers do 'nothing' to scale — the platform handles spikes automatically with on-demand overflow.
•20m watch time
Sort: