IBM Adds SSD Capability To Storage Virtualization Appliance

Controller storage virtualization

IBM's SVC, which was introduced in 2005, is a system of clustered appliances which sits in front of both IBM and non-IBM storage arrays to virtualize the storage into a single pool.

The addition of SSDs to the SVC improves the performance of all the attached virtualized storage, said Chris Saul, marketing manager for the SVC and storage virtualization at IBM.

Up to eight SVC nodes can be clustered together, with each node including one to four SSDs, for a maximum of 32 SSDs, Saul said.

When used with the SVC, the SSDs are a part of the primary storage capacity of the virtualized storage pool, and are not used as cache, he said.

"They can be used to increase the performance of the magnetic disks behind the SVC," he said. "And our Tivoli Productivity Center software can monitor the disks behind the SVC, see which applications can benefit most from running on an SSD, and automatically move those apps to the SSDs."

Instead of replacing spinning disk on a one-to-one basis, the SSDs are connected directly to the SVC's data bus for increased performance, Saul said.

IBM is using 146-GB SSDs from STEC, which has become the primary SSD supplier to major storage manufacturers.

In addition to adding SSDs, IBM tweaked the performance of the SVC in order to double the throughput compared to the previous generation. The company also tripled the maximum amount of cache per node to 24 GBs, and added 8-Gbit-per-second Fibre Channel connectivity.

Also new is the ability for the SVC to connect to both Fibre Channel and iSCSI SANs at the same time, Saul said.

The SVC's software was also tweaked so that it can now replicate data from applications such as Oracle, IBM's DB2, and Microsoft's SQL and Exchange without the need for scripts to coordinate the replication with the running of the application, Saul said.

IBM's SVC was already the strongest virtualization appliance not tied to a specific vendor's storage arrays, and with SSD IBM made it even better, said David Stone, vice president of business development at Solutions-II, a Littleton, Colo.-based solution provider and IBM partner.

"In that class of product, nothing can touch the SVC yet, and IBM just added years to the future of the SVC," Stone said.

Customers are not yet ordering SSDs by the terabytes, but they are noticing when vendors add SSDs that can be used to improve certain parts of their storage infrastructure, Stone said.

"We are already doing customer assessments for SSDs," he said. "Having SSDs in the SVC will help."

This is a great entry point for IBM into the solid state market, said Leif Morin, president of Key Information Systems, a solution provider and IBM partner.

"It lets them offer high input-output throughputs for random workloads, which is important especially for applications which need performance and have small data capacities," Morin said.

Solid state is getting to be a big deal, Morin said. "There are already applications for this offering," he said. "There's no doubt in my mind that random workloads will land on solid state technologies, especially as the price points fall."

The SVC 5.0 is expected to ship November 6 with a price starting at $40,000, which includes two SVC nodes, two UPSs, and the software. SSDs are priced starting at $24,999.