Abstract: Presented herein are techniques that automate data replication in a cluster by defining and managing logical availability zones. Nodes belonging to a same logical availability zone may be brought down together without impacting data availability. Thus, logical availability zones enable easier data management for events such as when nodes must be brought offline for troubleshooting or upgrading, while also providing resiliency against regional outages. Further, present embodiments define logical availability zones and assign nodes to zones in a manner that reduces the amount of data movement that is necessary, thereby requiring fewer computational resources.