Skip to Main Content


Question and Answer

Connor McDonald

Thanks for the question, Rene.

Asked: July 21, 2022 - 1:05 pm UTC

Last updated: August 04, 2022 - 11:42 pm UTC


Viewed 100+ times

You Asked

I have a client that is trying to implement KVM on an Exadata X9-2 M (, and he was told that in the Exadata he could have a single VM running in a single DB Node in the exadata, not a cluster VM, but just a single VM in a single DB Node.

To my understanding, that is not possible in the Exadata as they had to be VM clusters, 2 DB nodes, instead of 1 DB node.

I'm I incorrect in what I know?

and Connor said...

There are lots of things that can be done BUT not all of which are recommended / tested / supported / permitted by the standard deployment tools and so on

The best practice is to run Clusters, and RAC DBs, for availability and scaling. Also, the best practice is to use HIGH Redundancy. This means three copies of data (thus at three Storage Servers), and five ORC/Voting disks. Three of the five can the Storage Servers. When there are less than five Storage Servers, two DB nodes (Servers or VMs) can hold these, to make five.

With only one VM, and only three Storage Servers, you would not be able to have five, and thus would need to run with three. That could mean a Clusterware outage during a rolling update (one down for maintenance, and then a second down for an unplanned event.)

(Thanks to Steffen Weiberle for his input here.)