How Monzo tolerates full cloud outages
summary
Monzo builds an independent backup bank called Monzo Stand-in running in Google Cloud while the primary platform stays in AWS, with only 18 services and immutable synced state that produces deltas on an advice queue. Writes are about 2,000× lower than production, shadow tests and daily production enrolment validate behavior, and the system cost ~1% of cloud spend. A team of six delivers it in 10 months and it already reduces customer impact during a 2‑hour outage.
AI-written summary · about
chapters (11)
- 0:00Introduction to reliability
- 0:56The limits of cloud basics
- 2:36The challenges of multi-cloud
- 4:12Lessons from failure cases
- 5:38The Monzo Stand-in solution
- 8:34Data synchronization strategy
- 11:05State reconciliation process
- 12:02Testing in production
- 14:43Development and trade-offs
- 17:18Operational impact and value
- 19:20Lessons for other businesses
related talks
27mMigrating a Monolith to the Cloud
21mHybrid Cloud - nothing to worry about! 🌥️
35mBuilding a Bank with Kubernetes
45mZero production incidents. How to use an unpopular solution on a project
56mGetting all the 99.99(9) you always wanted
39mOops. I broke the Google, now what?
57mBuilding a bank from scratch in the Cloud
46mManaging a Billion Transactions with Tech, Tactics, and Teamwork