About cluster families
Cluster families group clusters that operate with a common purpose or under a common set of guidelines.
Clusters can be grouped in families to accomplish a particular task more efficiently or to imply a specific purpose within a grid. For example, one cluster family in a grid can be viewed as a production family. Another cluster family in the same grid can be viewed as an archive or as a family in the disaster recovery (DR) location. In either case, clusters within a common family perform certain grid-related tasks more effectively than clusters that do not share a family.
All members of a family employ Cooperative Replication and tape volume cache (TVC) Selection for Mounts enhancements. During a mount scenario, these enhancements improve control and performance by favoring cluster family members when remote mounts are required. The favoring of cluster family members is especially beneficial in distinguishing a set of production clusters from clusters intended for disaster recovery or archival purposes. Consistency across the family is the primary goal for cluster family members, and they collaborate to achieve this goal. Upon reaching family-wide consistency, the cluster family members then share data to bring each individual member up to the family consistency level.
The maximum number of possible families is equal to the maximum number of clusters in the grid. However, for cluster family benefits to be observed a cluster family must possess at least two cluster members. A cluster can belong only to one family at any time, but can change from one family to another. A cluster is not required to be a part of any family.
You can create, modify, or delete cluster families by using the TS7700 Management Interface.
Cooperative Replication
Cooperative Replication improves the efficiency and versatility of the TS7700's Copy Management function. Its enhancements to TVC selection enable the preferential selection of clusters within a cluster family when compared with clusters outside that family. As a result, clusters within the same cluster family can share the "Deferred" copy workload, optimizing throughput, and improving the overall time-to-family consistency.When a family becomes consistent regarding volumes outside of the family, the members of the family replicate among each other. This allows a family to become consistent at up to 1/n the time of volumes outside the family, where n is the number of clusters in the family. Since the clusters in a family are assumed to be at metro distances, the replication between them that later occurs is also accelerated.