Global Active Device: Remote, Virtualized Storage from HDS
When I wrote about HDS’s remote replication and mirroring technologies last month, I had left out another remote technology from the article, HDS’s Global Active Device. Global Active Device, supported by Hitachi’s VSP G1000, serves a separate but interesting purpose and I thought it deserved its own article separate from TrueCopy and Universal Replicator.
Virtualized Storage Machines
Global Active Device has key benefits built off of the concept of a virtual storage machine. For this virtual storage machine, at least three physical storage machines are needed. The first two, the primary and secondary storage systems, form the virtual storage machine which mirrors the information of the primary storage system. In this virtual storage machine, volumes are paired from the primary and secondary storage systems, as Global Active Device volumes. This pairing is synchronous, providing a remote copy of data to the secondary site. Additionally, the primary and secondary volumes are given identical LDEV numbers such that they will appear as just one LDEV to any host, and that the host will see the storage machine they are attached to as having the same information as the primary storage machine.
Behind the scenes, a quorum disk needs to be located on a third storage machine which will monitor the Global Active Device volume pairs. If a failure were to occur in the communication between the primary and secondary systems, they can refer to the quorum disk to diagnose the problem.
Seamless operation for hosts
The benefits of Global Active Device lie in fault tolerance, failover, and load balancing without storage access impact. Let’s cover what each of these mean from a host perspective.
Since volumes from two physically separate storage systems are involved in a Global Active Device pair, if one volume or system were to have a failure that prevents the host from accessing that volume, the host does not need to be concerned about a loss of access to the data on the volumes. Through allowing host I/O to continue by accessing the other volume at the other system, fault-tolerance is achieved at the storage level.
At the server level, failover can be enhanced through clustering software in which one or more servers have access to the Global Active Device volume through the virtualized storage machine such that if failure occurred on one server, failover to another server could be accomplished without any volume pair configuration actions needed in the process.
Finally, since the primary and secondary storage systems will hold the same data in each Global Active Device pair, load balancing can become simpler since there is no need to execute storage operations to prepare the data in a paired volume for use at the secondary site. The volume can already be seen by a second, remote host which could host, for example, a virtual machine which will ease the I/O burden from the primary storage system and primary host.
So, while Global Active Device isn’t quite the same idea as TrueCopy or Universal Replicator, I think that this is a great addition to HDS’s storage technologies. Even better, it has interoperability with some of the other replication software you can deploy on a VSP G1000, such as Thin Image, Shadow Image, and Universal Replicator.