Serverless - reality or BS - notes from the trenches
summary
Argues serverless wins big cost and speed gaps with concrete numbers and patterns: the Australian census fails after a $9M project while a serverless hackathon rebuild costs $500; BigQuery scans ~5 TB in 36s and reduces 84M rows to 2,500; Lambda often runs at roughly 1/100th the server cost and has a 1M‑request free tier. Shows AWS Lambda/S3/DynamoDB/API Gateway plus CloudWatch, Greengrass, Step Functions; Azure Functions/Logic Apps and Google Functions (beta); deployment via CloudFormation/SAM or Terraform; testing with mocha/chai and LambdaTester. Recommends one Lambda per verb, store state in S3/DynamoDB, and warming triggers for cold starts.
AI-written summary · about
chapters (23)
- 0:00Introduction
- 1:56Use case
- 3:42Definitions
- 4:10Amazon Lambda
- 5:13Amazon Lambda demo
- 8:16Serverless environment
- 8:44Microservices vs Serverless
- 10:33Cloud Services as Compute
- 11:50Serverless Definition
- 13:02Use Cases Patterns
- 15:06Demo
- 18:28Google Functions
- 22:25Azure Functions
- 24:22State of Serverless
- 26:42Google BigQuery
- 34:24Simple Viewer Service
- 36:42Lambdas
- 38:18Big Data
- 40:14Nordstrom Architecture
- 42:06Machine Learning
- 44:38Realworld problems
- 48:02Azure Logic Apps
- 55:14Monitoring
related talks
58mServerless - the brief past, the bewildering present, and the beautiful (?) future
42mNo Server Is Better Than No Server: Building Blocks for Serverless Architectures
54mServerless beyond Functions
42mAWS Lambda from the trenches
47mAdvanced Serverless Architectural Patterns on AWS
28mServerless - the way microservices should be
58mServerless: Five Key Things you need to Know
30mPerformance Tales of Serverless