FaaS or not to FaaS? Visible and invisible benefits of the Serverless paradigma
summary
Compares total cost of ownership for AWS serverless stacks — Lambda plus API Gateway, SNS/SQS/Kinesis, S3, DynamoDB, Aurora Serverless — and gives a decision checklist: use serverless for greenfield, event‑driven or asynchronous workloads; avoid it for GPU, sustained low‑latency or high‑throughput file processing. Calls out cold starts and mitigation via provisioned concurrency, VPC cold‑start improvements, Lambda limits (15‑minute max, 3 GB memory, account concurrency ~500–3000), RDS connection issues solvable with RDS Proxy/Aurora Data API, and higher API Gateway and DynamoDB on‑demand costs.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:42Value proposition
- 7:04Decision checklist
- 11:47Workload analysis
- 18:39Programming model
- 25:05Scaling and DB limits
- 29:40Cost management
- 34:34Organizational shift
- 39:24Tools and monitoring
- 45:32Future and conclusion
related talks
54mFaaS or not to FaaS — Visible and Invisible Advantages of the Serverless Paradigm
42mFaaS or not to FaaS: Visible & invisible benefits of the Serverless paradigm
01h 00mServerless - reality or BS - notes from the trenches
58mServerless Under the Hood
50mWhere We’re Going, We Don’t Need Servers!
01h 11mAWS RDS Aurora Serverless, ECS Fargate and Other Serverless Products
36mTransforming SaaS with Serverless Architectures
31mThe Evolution of Serverless: Past, Present and Future