Dancing with Serverless
summary
A Cloud Guru’s original platform runs as a serverless monolith that evolves into microservices: 289 Lambda functions, 13 services, 3.7 TB of S3 video, 107M Lambda invocations and 45M API requests in a month, scaling to ~650k then 2.2M users. Architecture uses Auth0, Stripe, Firebase (materialized view), Algolia, DynamoDB streams, GraphQL, CloudFormation/CDK and SST live Lambda development for local debugging. Observability uses CloudWatch/X‑Ray and Lumigo; orchestration uses Step Functions and EventBridge. Demonstrates serverless video transcoding by splitting, parallel Lambda processing and merge, often faster than single‑node encodes.
AI-written summary · about
chapters (16)
- 0:00Intro
- 3:27Early Days at ACG Teaching the world to cloud
- 6:51Evolution Teams of developers working in parallel
- 11:34Common Benefits When things go right
- 15:39Common Complaints Why can't things just be easy
- 18:02How I build today Back to working on my own
- 20:26Serverless Stack (SST)
- 24:13Observability
- 25:15Fatfire Architecture
- 31:07Web Sockets With API Gateway
- 37:18Step Functions Coordinate Distributed Applications
- 38:46Parallel Architecture Dealing with unexpected problems
- 40:19Divide and conquer Using the Lambda supercomputer
- 41:22Parallel Computing with Lambda & Step Functions
- 43:20Serverless Video Transcoder Parallel and conquer
- 48:36Ampt
related talks
39mServerless in production, an experience report
29mHow to build a social network on serverless
31mAccelerate business innovation with serverless applications
01h 02mServerless in production, an experience report
41mTechnology mix: Around the serverless in 180 days
13mSam Kroonenburg, A Cloud Guru | ServerlessConf 2018
42mAWS Lambda from the trenches
42mNo Server Is Better Than No Server: Building Blocks for Serverless Architectures