PowerHA SystemMirror for AIX Enterprise Edition
Increasingly, IT organizations are being asked to provide multi-site configurations for disaster recovery as well as to periodically demonstrate compliance to corporate policy and regulations. In the past, this was typically a recover from tape operation or a reboot operation at the remote location. The PowerHA SystemMirror Enterprise Edition is essential for clients who must both protect themselves from site-wide failures as well as to be able to readily demonstrate the capability to do so. PowerHA V7 enables clients to choose an operator managed failover policy for multi-site linked clusters. In this mode, the operator is the final arbiter deciding whether or not to failover to the alternate site.
Storage servers or host based replication provide the data replication services and the Enterprise Edition provides management and automation of the failover process. PowerHA SystemMirror Enterprise Edition V7 supports HyperSwap configurations with the IBM DS8800 or DS8870 in a Metro Mirror configuration. This advanced technology provides the capability for a cluster to span two sites with the storage and servers cross coupled in a manner that keeps the application resilient through either a storage server or a production server outage (for active-active workloads). HyperSwap can also be deployed in single AIX LPAR configurations. Customers can enable HyperSwap for multi node cluster configurations while at the same time exploiting the storage mirroring and swapping capabilities for single node AIX LPARs not in the cluster.
Geographically dispersed configurations can be deployed via a stretched cluster configuration defined as having a single cluster repository or via a linked cluster configuration with two independent yet linked cluster repositories. A stretch cluster can also be deployed with the Standard Edition and LVM mirroring in a cross site mirror configuration similar to what clients have deployed in previous versions. The stretched cluster configuration provides three levels of cluster communication redundancy via multi-cast across the network, SAN fabric and repository disk. Multicast network communication requires that switches on the network fabric connecting the hosts allow multicast IP packets to flow through them. PowerHA V7 also supports clustering based on non-multicast (unicast) IP communication for data center deployments.