Overview
Explore common mistakes and misconceptions when working with AWS Lambda in this GeeCON Prague 2022 conference talk. Dive into the intricacies of serverless computing, focusing on AWS Lambda's flexibility and integration capabilities. Learn about the challenges that arise as client numbers and requests increase, including latency issues and unexpected costs. Discover strategies to mitigate cold starts, optimize JVM performance, and effectively use provisioned concurrency. Compare Lambda pricing with other platforms like EC2 and Fargate, and understand how to calculate AWS costs accurately. Examine real-world scenarios, from simple "Hello World" applications to simulations of high-traffic systems like Stack Exchange. Address the unique challenges of caching in Lambda environments and explore various approaches to overcome them. Gain valuable insights to enhance your serverless architecture and avoid pitfalls in AWS Lambda implementations.
Syllabus
Intro
Lambda instance lifecycle
Cold start type 1 (container recycle)
Cold start type 2 (JVM JIT and lazy load)
Reducing cold starts
Cold start for provisioned concurrency
Provisioned concurrency best practices
Lambda pricing
How to calculate AWS costs
Assumptions
Scenarios considered
"Hello world", 1 request per second
Fargate, 1 request per second
Simulating Stack Exchange
Caching in Lambda - challenges
Caching in Lambda - approaches
Taught by
GeeCON Conference