Migrating heavier data loads has gotten cheaper — and less painful — in recent years, as companies harness the power of virtualization. This week, Hitachi Data Systems
announced a service that it said was a first in the industry: data migration without any host and application downtime.
The storage giant says it can now migrate customer data from its Universal Storage Platforms (USPs) to its Virtual Storage Platforms (VSPs) without any outage or impact on server performance. The technology used is based on using initiator and target machines and spoofing the SCSI IDs of source logical devices. It can also be performed across multiple platforms, the company says.
“It allows organizations to completely eliminate the outage window and business disruption, which is part of the promise of virtualization,” says Chris Willis, Americas director of cloud and converged solutions at HDS. “Once virtualized, there are no migrations with downtime.”
David Russell, research vice president of storage technologies and strategies at Gartner Inc
., says the term “non-disruptive” has been a bit nebulous in the storage world in the past. “Going back well over 10 years ago, they would talk about ‘non-distruptive migration,’ but that was after you took a power outage, deployed new cards in the servers
, installed new software, and re-cabled the network. And from that point on, you could do non-disruptive migration. But that weekend was fairly disruptive.”
But HDS's new capabilities, says Willis, allow migration to proceed “while the application heart is beating.”
“We joke, but there’s only two professions [in which] you actually operate on the patient while they’re conscious. One is neurosurgery and the other one is storage.”
Large enterprises will be particularly interested in technology that allows them to virtualize their storage and migrate it faster and more efficiently, adds Russell. He says the more than $15,000 per-terabyte average figure that Hitachi estimated for enterprise migration costs is not far off when taking into account the Tier 1 storage, “the infrastructure, the people, the network connectivity, the amount of time.”
He says the “nightmare scenario” for companies is when the leases or warranties on multiple large storage arrays expire at the same time.“That’s really very invasive, because it’s not just all of that storage, but it’s also going up the stack: what are all the business processes, the applications, the administrative interaction that might have to get coordinated to be able to achieve that?”
Automating your storage migration to the extent VSP can achieve also translates into cost savings by handling some of the work a collection of IT people would have to be paid for, adds Willis.
“The typical migration involves having database administrators on site, it involves having application folks on site. It involves having the UNIX and storage administators on site. By the time you load up the entire room with technology professionals, the cost per hour to migrate a terabyte is quite high.”
Russell says performing any simultaneous migration is quite the technical feat. It involves using state-of-the-art technology that has only recently become available. “You’re looking to physically move blocks of data while maintaining access to them externally, to different applications, different hosts or machines. It’s not an easy trick.”
“It would be like trying to continue to watch television on one channel while simultaneously turning the tuner. We’ve seen things like digital video recorders that can actually do some of that, allow you to record one program while you watch another, but it turns out that it wasn’t too many years ago when that wouldn’t have been possible.”