Two Years Later, Loving and Hati
summary
Picnik runs a hybrid stack: a small Seattle cage plus AWS S3 and EC2, using render jobs on EC2 and queue-driven scaling for peak load while repurposing virtualized cores for web vs. compute. Storage hit scale problems—nearly 1 billion S3 objects and a policy that pushed 98% of files to S3—and a deletion backlog that balloons costs until ops intervenes. Monitoring uses smokeping and BGP route changes to work around cross-region network failures; a site-level “offline percent” forces graceful shutdown when EC2 capacity disappears.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 0:24Picnic and cloud journey
- 1:47Hybrid infrastructure
- 3:35Storage with S3
- 6:22Hybrid cloud lessons
- 7:25Distributed cloud challenges
- 9:54Cloud flexibility and thoughts
related talks
46mHighly Scalable Image Storage with AWS Serverless
38mThe SmugMug Tale
21mThe best compute for your storage - Amazon S3 & AWS Lambda
27mScalable by Design: How to Build an Elastic Cloud Infrastructure...
28mCloud The Final Destination
51mArchitecting for Hyperscale with AWS: Navigating Real-world Serverless Challenges
27mThe journey from legacy to cloud - native success
01h 03mWindows at Scale