Temporal at Scale: Lessons in Migration and Security
summary
A 12-person team coordinates a zero-downtime migration of 300+ namespaces and services from a custom self-hosted Temporal stack to Temporal Cloud, reusing an Aurora staging phase to prove patterns. Migration tooling includes a multi-client manager with percent-based routing and dual workers for no forced termination, plus drain-and-switch for short-lived workflows. Security uses private links, mTLS plus certificate filters, and an optional centralized codec server for payload encryption. Observability, canaries, and rollback paths keep the cutovers safe and reversible.
AI-written summary · about
chapters (13)
- 0:00Introduction and speakers
- 1:27What is Coinbase and Temporal
- 3:12Why migrate to Temporal Cloud
- 6:45On-prem pain points
- 8:13Security model for Temporal
- 13:05Building trust with teams
- 14:50Observability during migration
- 15:54Migration journey overview
- 17:40Drain and switch strategy
- 20:51Dual workers migration
- 26:25Custom execution strategies
- 30:38Lessons and takeaways
- 31:41Closing and acknowledgments
related talks
01h 03mContinuous Security in the Cloud
28mSurviving migration madness - Humana's move from Terraform Enterprise to Terraform Cloud
39mLessons Learned Migrating an Existing Product to a Multi Tenant Cloud Native Environment
15mScaling AI Agents Without Breaking Reliability
25mMass migrations to the cloud
21mEscape Plan: How We Survived Without EKS and Learned To Love Air-Gapped
45mTemporary Access to the Cloud: A Case Study
30mBeating the competition through faster time to market