1

I have a general architectural decision question relating to AWS RDS. We have an Oracle RDS DB for our application. It is a Single AZ DB with Daily Snapshots. That is quite enough for us as we are not a High Availability system. In case of failiure we have almost 5 days time to restore the DB, i.e. we are not in a rush.

However what would happen in a catastrophic scenario, lets say a bomb destroys one or more Data Centers in one AZ and that one Availability Zone is unavailable, and would probably remain so for some time. I know its an unrelisitic scenario but it is something our customer wants to be safe from so it is a real requirement.

Naturally we would restore the DB from a snapshot in another AZ. However the question is if we wouild be able to do so, i.e. if AWS would have the capacity to do so, as if an AZ is suddenly unavailable many customer would move their systems to the other 2 available AZs and it can happen that we wouldn't be able to get the necesarry hardware to start an instance there. At least that is our reasoning.

Is our reasoning justified or are we missing smth? How could we guarantee that we get an instance in the remaining unaffected AZs?

Thanks

vlatkovr
  • 19
  • 3
  • AWS offers [capacity reservations](https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-capacity-reservations.html). Thus, if you really need to ensure space in AZ for your app, you can buy the reservations. – Marcin Aug 28 '20 at 05:55
  • Thanks @Marcin. That is what I would need but it is only for EC2 Instances. The Problem I have is with an RDS DB. – vlatkovr Aug 28 '20 at 06:43
  • There are [reserved instances](https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/USER_WorkingWithReservedDBInstances.html) for rds, but they don't provide capacity reservation as well. – Marcin Aug 28 '20 at 06:52

0 Answers0