Look Ma, no OPS! Automating microservices' infrastructure
summary
Team automates AWS infra for dozens of microservices using Terraform instead of CloudFormation, scaling from a few dozen to hundreds of resources across environments and removing ops bottlenecks. They build a Gradle plugin installs Terraform and runs init/plan/apply, enforce remote state in S3 with KMS encryption and state locking, and use modules from Git with version tags. The flow includes automated terraform binary management, plan review, controlled apply, import of existing resources and destroy tests, and it proves to work in production.
AI-written summary · about
chapters (19)
- 0:00Introduction and challenges
- 2:12Limitations of CloudFormation
- 3:10Introduction to Terraform
- 5:46Managing terraform state
- 7:45Goals of the new solution
- 9:30Initial naive Gradle approach
- 10:38Developing a Gradle plugin
- 12:14Automating terraform install
- 13:35First demo
- 17:00Terraform modules
- 19:07Remote backends and locking
- 21:18Refactoring the build flow
- 23:15Gradle task dependencies
- 24:31Second demo
- 28:30Third demo
- 33:33Addressing human resistance
- 36:57Final demo
- 37:38Best practices and conclusion
- 43:46Q&A
related talks
01h 32mUsing Terraform and AWS to create environments on the fly
35mMisadventures with Terraform
52mTest Driven Delivery of Infrastructure-As-Code
55mSelf-service infrastructure is no longer a dream
29mInfinity and beyond: Scaling Terraform for the global enterprise
44mTaming the cloud sprawl
15mTerraform, GitOps & Kubernetes: manage infrastructure & aas
29mBreaking Down the TofuLith. How We Reduced Lead Times for Multi-Tenant Infra Changes...