This article first appeared in "Storage" magazine in the May issue. For more articles of this type, please visit www.storagemagazine.com.
At issue: New virtualization products offer more than unified access to storage.
Virtualization is hip again, but this time around it has a new twist. It's no longer about delivering a single pool of storage across heterogeneous storage, like it was a couple of years ago. These days, virtualization is all about delivering data services, as in replication, mirroring and data migration.
In April at Storage Networking World, Hitachi Data Systems (HDS) and Sun Microsystems announced new copy and data movement services for their respective virtualization platforms. HDS announced HiCommand Tiered Storage Manager for the TagmaStore Universal Storage Platform, which allows users to do policy-based data movement between storage resources under TagmaStore's control. Sun, meanwhile, announced an enhanced version of its 6920 platform, which in addition to support for EMC, Hewlett-Packard and IBM arrays, can also be outfitted with the StorEdge Pool Manager, Data Replicator, Data Mirror and Data Snapshot services.
In May, EMC announced its own virtualization platform, called Invista. It, too, EMC says, will be enlisted in the service of data services -- data migration, to be specific.
The funny thing about these virtualization platforms is that, more often than not, customers don't actually use their virtualization capabilities, says Hu Yoshida, HDS' CTO. "We can either remap the data through the [TagmaStore] controller, or we can do a 1:1 mapping," he says, whereby the virtualization controller leaves the LUN's original data descriptors in place. "Most people have gone with the 1:1 mapping." HDS sometimes refers to this capability as "native mode." IBM SAN Volume Controller customers know it as "image mode."
Even when the data isn't virtualized per se, Yoshida says, "we're still able to apply all our functionality," such as replication and data movement. Data movement, in particular, happens for three basic reasons, he says: information lifecycle management, performance and "technology refresh."
Heterogeneous storage pooling, meanwhile, is still quite rare, says John McArthur, IDC storage analyst. There are cases of it, but for now, "it's really more a data migration play."
For more information:
About the author: Alex Barrett is "Storage" magazine's trends editor.