Where a CR or referenced configuration is changed, the handling of those changes should be done in a roll-out manner
For example, a simplistic approach to adding a server to a platform would be to restart the platform. However this would lead to an outage of the other servers on that platform.
If instead platforms are bounced one by one until all are complete, then no outage will occur (unless only one replica is in use)
Some changes may not require any outage, and could therefore be done across all replicas in short succession.
Where a CR or referenced configuration is changed, the handling of those changes should be done in a roll-out manner
For example, a simplistic approach to adding a server to a platform would be to restart the platform. However this would lead to an outage of the other servers on that platform.
If instead platforms are bounced one by one until all are complete, then no outage will occur (unless only one replica is in use)
Some changes may not require any outage, and could therefore be done across all replicas in short succession.