Working with AWS Lambda without Breaking the Bank
summary
Explains concrete cost levers for AWS Lambda: billing equals requests × duration, memory between 128 MB and 10 GB controls CPU and network, and increasing memory can shorten duration; shows how runaway recursive invocations and repeated cold starts inflate bills and how connection reuse and dead‑letter queues prevent retries. Demonstrates using CloudWatch metrics and Lambda Power Tuning plus Compute Optimizer to find memory sweet spots, contrasts reserved concurrency vs provisioned concurrency with a 400‑slot example, and gives monitoring+throttling patterns that cut real invoices.
AI-written summary · about
chapters (6)
- 0:00Introduction and speaker intro
- 4:13Lambda basics and cost factors
- 12:22Millionaire game setup
- 22:33Optimizing Lambda performance
- 38:07Cost optimization best practices
- 39:34Q&A and closing
related talks
54mContainerless in the cloud with AWS Lambda
33mAWS Lambda - Logic in the Cloud without Back-end
39mThink in Cloud, Develop in Cloud
42mBest practices for serverless developers
54mHow to avoid common mistakes and misconceptions when working...
21mBeyond the Basics: Production Serverless Patterns for Extreme Scale
30mAnd the CFO Wept: AWS Cost Control
38mAll you need to know about AWS: Lambda