Multi-Region Is Not a Backup Plan
Pillar 03 · Residency
Every eighteen months or so, a data centre in northern Virginia has a bad morning, and a few million businesses discover they were all renting from the same landlord. Banks, airlines, baby monitors, somebody's smart fridge. All down. All surprised.
Then comes the status page. Yellow, then orange, then the little note that says "we are aware of an issue affecting some customers," where "some" means "everyone" and "an issue" means "the building is on fire."
And the post-mortems all say the same brave thing. "We had multi-region redundancy." Congratulations. You had two apartments in the same building, wired to the same breaker, insured by the same company. The redundancy was theatre and the bill was real.
Here's what nobody mentions when they sell you "multi-region." The regions share a control plane. They share an identity system. They lean on a handful of global services — DNS, auth, the metadata endpoint — that have no second copy at all. So when the thing that fails is the thing underneath all the regions, your failover region fails too, in perfect synchrony, because it was never independent. It just had a different postal code.
That is not resilience. That is the same single point of failure, billed twice, with a diagram that looks reassuring in the procurement deck.
Resilience means surviving the loss of an entire provider. Not a region. Not an availability zone. The whole vendor, control plane and all, gone for the afternoon.
CloudRAID erasure-codes your data into shards and spreads them across independent providers, so any sufficient subset rebuilds the whole. One cloud goes dark, you reconstruct from the ones still standing. No failover runbook nobody tested. No war room. No carefully worded apology explaining to your customers that "the cloud" was down, as though the cloud were weather and not a building you chose.
The design doesn't care which vendor is having the bad morning. It assumes one of them always is. That assumption has not once been wrong.
You can keep buying two copies of the same outage. Or you can stop subscribing to it.
And yes — because the shards land in different jurisdictions, sovereignty falls out for free. That's the footnote. Staying up was the point.
More from the blog
CBC and CTV Say Canada's Cloud Market Is "Broken." They're Half Right.
A Better Question Doesn't Survive a Subpoena
Bergson Lopes Rego published a piece in CDO Magazine called "The Data Sovereignty Illusion." Read it. The diagnosis is…
Read → Commentary · 5 min readHave You Ever Wondered Where Your Data Goes in the Cloud?
You upload the quarterly numbers. A little spinner turns. "Saved to the cloud." Reassuring phrase, the cloud. Sounds…
Read → Regulation · 3 min readThe Kill Switch Has a Loyalty Program - Microsoft is in the Trump trap
Three weeks before Brad Smith promised Europe that Microsoft would protect it from Washington, Microsoft had already…
Read →