From Laptop to Lambda: Outsourcing Everyday Jobs to Thousands of Transient Functional Containers
summary
A toolkit called gg packages programs as content‑hashed lightweight containers called gg thunks and composes them into a linked GG IR to offload entire jobs to AWS Lambda, avoiding laptop round trips. It uses HTTP pipelining and direct worker-to-worker transfers (observed up to 600 Mb/s) to move data. Results: distributed compilation outperforms icecc by 2–5×, stays within 20% of ExCamera for video, compiles Chrome in ~18 minutes, and an FFmpeg demo uploads 100 MB in 6s and finishes for ~$0.30.
AI-written summary · about
chapters (7)
- 0:00The challenge of outsourcing
- 2:07Managing dependencies
- 4:59Reducing round-trip latency
- 8:22Scaling in the cloud
- 11:13Automatic build integration
- 15:30Performance results
- 17:02Q&A session
related talks
39mHow to Go with AWS Lambda?
23mOn-demand Container Loading in AWS Lambda
47mServerless Compute at the Heart of Your EDA
30mTerrible Ideas In Lambda
33mContainers and AWS Lambda: Choose the optimal compute for applications
48mServerless beyond functions
52mGo and AWS Lambda – a perfect Match
40mServerless Compute at the Heart of Your EDA