NEW YORK -- Network Appliance Inc. has outlined a strategy for migrating its user base to a new storage grid architecture...
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
based on the technology it acquired a year ago from Spinnaker Networks Inc., but the crossover period could be painful and off-putting, some analysts think.
At a media and analyst event in New York, NetApp said its three-phase roadmap for introducing the grid-based operating system will be transparent to users as it plans to take features from its existing Data ONTAP OS and incorporate them into the SpinServer product.
"The aim is not to disrupt ONTAP customers," said Dave Hitz, VP and co-founder of Network Appliance. "Think of the upgrade like moving from Windows to Windows NT. There's a long overlap period … All future development is focused on the next-generation product, but we'll support both."
The first phase of the integration dubbed storage convergence is complete today. RAID DP is available for SpinOS, for example. Phase two, which NetApp is calling data management convergence, involves integrating all its applications like SnapShot and SnapMirror into the Spinnaker operating system. The target date for this effort is the end of 2005. And the final phase, or full convergence, when the product will support block storage and iSCSI protocols is targeted for the end of 2006.
In the meantime, NetApp users who want the SpinServer must invest in and manage two separate hardware platforms and two separate operating systems. "That's not trivial -- you've got separate device drivers, separate management tools, separate ways to set up RAID groups for WAFL than for SpinOS … I don't think users are going to buy into it until it's a merged product," said one analyst who requested anonymity.
Brian Babineau, analyst with Enterprise Strategy Group, noted that NetApp will not be pushing the SpinServer until the convergence process takes place. "They are not selling Spinnaker right now -- you've got to really ask for it. There are a few lab rats using it today but that's it."
SpinServer solves a scalability problem that has begun to plague network storage as more and more of it gets deployed. Until this product and others like it, NAS and SAN storage has been implemented in islands, servicing only the applications it was originally targeted for. This works fine in small doses, but as users add more and more storage, it becomes harder to manage and utilize efficiently. SpinServer offers a way around this using a distributed file system and virtualization to pool all the storage together so that it can be managed as a single, unified system. Then migrating files from one SpinServer to another, for cost purposes or to free up resources, should be as simple as clicking a mouse.
Ron Bianchini, vice president and general manager of scalable systems at Network Appliance, and formerly the CEO of Spinnaker, said the initial target for the product was high-performance computing, because "they had the most pain." But he believes it will eventually be adopted by the mainstream. There are between 40 and 50 installations of SpinServer in the market today, but only a small percentage of these are running their core datacenter applications on it, Bianchini acknowledged.
Laszio Rausch, VP of technology at ABN Ambro, a global bank with $107 billion in assets, has about 400 TBs of NetApp storage but won't be using the SpinServer for file migration or volume management. "We want this capability to be independent of the storage, so we are using Cisco's MDS switches for a higher level of flexibility," he said. Rausch will use SpinServer to scale to larger volumes and to manage larger numbers of clusters.
Chad Haywood, principal at State Street Bank, looked at the Spinnaker product before Network Appliance acquired the company but ruled it out because of the lack of synchronous replication. Instead State Street uses EMC's SRDF (synchronous remote data facility) software to replicate data between Boston and Westborough, Mass., for disaster recovery purposes.