Terraforming RDS: What Instacart has learned managing over 50 AWS RDS Postgres instances with Terr
summary
Instacart manages ~80 AWS RDS Postgres instances with Terraform and recounts terraforming ~15,000 AWS resources; the talk shows concrete failures and fixes: Terraform/AWS defaults hide values, apply immediately set false causes storage increases to not take effect, a unit change in max_wal_size (from 16MB units to 1MB) forces excessive checkpoints and CPU spikes, and parameter group plans are noisy. Includes modules for RDS, EC2 PGBouncers, lessons on when modules hurt versus when Terraform wins.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:48RDS and Terraform pain points
- 8:13Apply immediately pitfalls
- 11:06Parameter group challenges
- 17:07Real-world upgrade disaster
- 21:02Module design lessons
- 26:01Why Terraform still wins
related talks
01h 03mTerraforming the world, one cloud at a time.
58mAchieving unparalleled elasticity on AWS at the lowest cost
52mBuilding Database Using Terraform in AWS for Maximal Isolation Usability
30mTerraform in Depth QA and Fireside Chat
41mTerraform Practices - The Good, the Bad, and the Ugly
44mTaming the cloud sprawl
30mTerraform Practices to Enable Infrastructure Scaling
31mTerraform Practices to Enable Infrastructure Scaling