Suhail Patel at StaffPlus London 2023
summary
Argument: run migrations continuously, not as one-off projects. Concrete numbers and tools: monzo runs 2500 microservices and deploys 100+ times/day, moves Cassandra to Amazon Keyspaces using an automated migration service that implements the runbook as a linear state machine, uses shadow traffic and Semgrep static analysis to block unsupported code. Pattern: centralized migration team, tiered rollout from low to critical services, coherence guinea‑pig services for continuous tests, automated notifications to owners. Results: automation repeatedly catches failures and makes migrations repeatable.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:03Running migrations necessity and risk
- 2:49Monzo’s platform team philosophy
- 4:55TSB migration disaster lessons
- 6:20Pitfalls of one-off migrations
- 9:33Breaking production cautionary tale
- 11:17Migration time experiments and support
- 13:46Shadow mode and non-disruptive testing
- 14:30Architecture reviews as collaboration
- 17:38Proposals versus reality in migrations
- 18:40Centralized vs decentralized migrations
- 20:28Detailed runbooks for safe migrations
- 21:50Pre-mortems and automated checks
- 24:41Static analysis for migration rules
- 26:04Automating migrations with service
- 28:10Tiered migration for critical services
- 30:36Incidents and trust during migrations
- 32:20Celebrating milestones and retrospectives
related talks
55mMigrate your company to microservices: a step by step tutorial
28mSaaS migration and modernisation
30mArgo Rollouts at Scale: How We Brought Automated Rollback to 2100+ Micro-serv...
01h 00mSix Secrets to Succeeding with Microservices
42mInnovating Faster with Micro-Service Architecture...
52mTesting Microservices
30mMonolith to Microservices
21mThe great SOA migration