Trick or Treat Your Pipeline: A Haunting Tale...
summary
Company runs 1,200 Lambda functions in production and uses CircleCI to push artifacts into a shared builds bucket, then triggers environment-specific deploy agents (the in-house Buzaglo deploy agent) via S3→SNS→SQS so CI stays environment-agnostic. Engineers use CDK/CloudFormation for service stacks while DevOps manages stateful infra with Terraform. Production deploys go through API Gateway + IAM for gating. Deployments are immutable builds so tagging stacks enables immediate rollbacks and per-deployment monitoring in Datadog.
AI-written summary · about
chapters (10)
- 0:00Introduction and Milo overview
- 10:23Serverless CI/CD pipeline basics
- 12:26Immutable builds and deployment issues
- 18:14Decoupling CI from environments
- 20:57Production deployment with API Gateway
- 24:13Monitoring with tagged deployments
- 29:50Rollback using immutable builds
- 31:16Buzaglo deployment agent explained
- 36:11Key takeaways and best practices
- 39:46Q&A and closing remarks
related talks
35mBeyond build/test/deploy
26mIntelligent Deployment Pipelines
39mBuilding Serverless Application Pipelines
56mFast & Simple Deployments for Code & Infra
23mCI/CD at Scale: Best practices with AWS DevOps Services
30mPipeline patterns to aid your next deployment
01h 03mYet Another Continuous Delivery Talk
22mA Crawl, Walk, Run Story of CI/CD