Serverless in production, an experience report
summary
Company migrates a monolith to an event-driven, service-oriented system centered on AWS Lambda and ships concrete wins: 170 Lambdas in production, running at about 5% of EC2 cost and boosting releases from 4–6/month to ~100 releases/month. Built realtime search (events → Kinesis → Lambda → CloudSearch), analytics (Kinesis → Lambda → Google BigQuery), a feed pipeline (events → SNS → Lambda → Redis), hourly trending via BigQuery → DynamoDB, CI/CD with Serverless framework, centralized logging, tracing, Parameter Store for config, and a janitor Lambda to avoid the 75 GB package limit.
AI-written summary · about
chapters (9)
- 0:00Start
- 21:36Levels to Testing
- 22:22Integration Tests
- 32:37Whether To Do Continuous Delivery or Continues Deployments
- 33:25Logging
- 34:38Centralized Logging
- 37:56Correlation Ids
- 49:01Parameter Store
- 50:39Listening to Signals
related talks
39mServerless in production, an experience report
47mServerless in production, an experience report
01h 02mFrom Monolith to Serverless
01h 04mMigrating to Serverless - an experience report
42mAWS Lambda from the trenches
29mHow to build a social network on serverless
42mEnterprise Serverless Adoption. An Experience Report
45mPatterns & practices for developing enterprise-scale serverless applications
