Faster, Cheaper, Greener: Pick Three
summary
Rearchitected a monolithic rule engine into a cloud native, share‑nothing pipeline on AWS using Step Functions, SQS, S3, DynamoDB, Lambda, Fargate/ECS and Aurora Serverless v2. The distributed system spins 50–300 Fargate containers per client, migrates 90% of clients off the old stack, and solves noisy‑neighbor CPU bottlenecks. One large client’s run drops from 1h40 to 39 minutes, relational storage pressure falls (one DB had 21 TB with 19 TB JSON) and costs drop ~70% for DB storage.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:26Sustainability in software matters
- 3:25Shareholding disclosure challenges
- 11:39Old monolithic architecture problems
- 17:09Rebuilding rule engine cloud-native
- 22:11Distributing compute with Fargate
- 26:23Offloading storage to S3
- 31:23Aurora Serverless for results
- 36:22Benefits of new architecture
- 43:23Cloud providers greener
- 47:13Steps to greener software
- 51:10Code efficiency limits
- 56:28Urgent call to green software
- 58:19Q&A
related talks
23mRun Less Software; Use Less Bits
27mThe journey from legacy to cloud - native success
46mArchitecting for Success and Runway
18mThe nonstop way to find AWS-recommended savings opportunities
38mRefactoring to serverless
37mBuilding Modern Distributed Systems with AWS
49mBuilding Next-Generation Applications with EDA
51mArchitecting for Hyperscale with AWS: Navigating Real-world Serverless Challenges