What would happen to our data if our GCP Datacenter goes down?

Following the recent events in France at OVHCloud's Strasbourg Datacenter, we can't help but wonder what would happen to our whole stack in the cloud if the DC we are currently hosted on at GCP meets a similar fate.

We have everything running through k8s on 3-4 different VM's.

Is there a fragmentation of the data across multiple DC's of a region that would prevent a total loss of data ? Is there some last resort plan at their DC's that would for example start migrating everything to another one of their DC's if anything dangerous starts happening ?

The underlying question here is : would we need to properly duplicate our services across at least 2 regions to be certain that this could never happen to us ?

Solution 1:

Google has different types of storage and different tiers of those types.

If you have your data on local disks (https://cloud.google.com/local-ssd), you would be lost.

With persistent disks, it depends. Except regional PDs (https://cloud.google.com/compute/docs/disks/high-availability-regional-persistent-disk), (per SLA) all PDs are only in one zone in one region.

You can find an overview over all current options on https://cloud.google.com/compute/docs/disks

You can store snapshots across multiple regions: https://cloud.google.com/compute/docs/disks/snapshots

When creating you K8S cluster, you can configure it to be regional. In that case by default that cluster runs in 3 zones of a region: https://cloud.google.com/kubernetes-engine/docs/concepts/regional-clusters

A final statement: What happend with OVH is very unlikely to happen with Google or AWS (and I guess also unlikely with Azure). But other things may happen and have happend. So in any case, if availability and time to recover in case of a disaster is one of your goals, you should store your data and run your application in at least two regions of at least two independent Clouds.