When it comes to cloud-based disaster recovery, one of the most important things to remember is that there is no...
such thing as 100% reliability. Cloud service providers spend huge amounts of money to implement redundancy on every level of their infrastructure, yet we still hear about cloud outages. The best cloud-based option is a multi-cloud disaster recovery strategy.
Major public clouds tend to be just as reliable -- perhaps more so -- as any on-premises infrastructure, but using multiple clouds is one way to ensure a higher level of protection.
This option can be a little bit tougher to implement, because each cloud provider has different standards for its virtual machines. Even so, a multi-cloud disaster recovery strategy provides another layer of security.
A major problem with disaster recovery planning is that you never know exactly what a disaster might entail or how widespread the effects of the disaster might be. Under "normal" conditions, the chances of a cloud provider experiencing a major outage at the same time that your organization is dealing with a disaster are near zero. However, a large-scale disaster could conceivably impact both your organization and the cloud that you use for disaster recovery. A major storm for example, can span many hundreds of miles, and if a cloud provider's data center is in close enough proximity to your own, then you might both be impacted by the same storm.
If you opt to use a single cloud, don't keep all of your resources in the same region. It's fine to replicate mission-critical resources to the cloud region that is geographically closest to you, but, ideally, you should be using multi-cloud disaster recovery to replicate your resources to a second cloud region further away.
Why bother with multiple clouds if you can already replicate to multiple regions with just one? With one cloud, the data centers are all part of that same cloud. If the cloud provider were to experience some sort of global failure, then all of its regions could conceivably be impacted. On the other hand, there is almost no chance that two different cloud providers will both experience a catastrophic global outage at the same time.
If you opt to use a multi-cloud disaster recovery strategy as part of your cloud-based DR plan, then you should try to avoid overlapping regions. If, for example, you use one provider's east coast United States region, then you should use a different region with your other cloud provider. That way, a regional disaster can't take down both of your clouds at the same time.