Breaking Down the TofuLith. How We Reduced Lead Times for Multi-Tenant Infra Changes...
summary
Claims a 90% lead time reduction by splitting a monolithic Terraform state into per-product workspaces, moving to per-tenant AWS accounts, replacing the incumbent with Scalr to lift limits to about 120 concurrent runs, and creating repeatable modules plus a config-driven vending machine. State splits grow to 8,453 workspaces, active Terraform contributors rise from ~8 to ~45, ops team dissolves into product teams, training and API guardrails prevent risky changes (eg blocking >20–30% deletions) and allow many engineers to apply infra safely.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:05Infrastructure team as constraint
- 3:30Multi-tenant AWS accounts
- 5:17Scalr for concurrency
- 7:45Terraform state workspaces
- 10:09Consolidating vending machines
- 12:37Dependencies in workspaces
- 14:21Training for Terraform
- 16:51Dismantling ops team
- 18:18Workspaces state today
- 20:47Success challenges
- 22:3390% lead time reduction
- 23:14Q&A
related talks
11mLightning Talk: Migrating a Client onto OpenTofu for Cost + Speed
44mTaming the cloud sprawl
29mWhat would you do with 8k extra hours? Turning infrastructure into a strategic advantage
27mCase Study: Re-Thinking Our Infrastructure Tooling
57mLook Ma, no OPS! Automating microservices' infrastructure
22mCase Study: Re-Thinking Our Infrastructure Tooling
58mAchieving unparalleled elasticity on AWS at the lowest cost
31mTerraform Practices to Enable Infrastructure Scaling