By Heidi Biggar
The story Jim Livingston, technology architect and director of the Data Resource Center for the University of Utah Health Sciences Center (UUHSC), recounts is one that many IT administrators-particularly those at medium-sized and large enterprises-can relate to.
While storage networking has solved some storage problems (e.g., issues with data sharing, performance, and LAN traffic), it has created others. In the case of UUHSC, for example, it did little, if anything, to address the university’s growing “stranded storage” problem. In fact, according to Livingston, it merely shifted the problem from one environment to the next. (UUHSC has been gradually migrating from direct-attached storage, or DAS, to SANs.)
“Even in a SAN environment, you’ve still got stranded capacity to deal with,” Livingston explains. “You’re provisioning capacity in huge chunks that can take a year or two for applications to grow into [and] that can’t be utilized by all servers. So all you’re doing is migrating the stranded capacity problem from the DAS environment to the SAN.”
As more applications were brought online, UUHSC’s problems worsened: Large chunks of capacity were doled out indiscriminately to applications regardless of need, resulting in islands of underutilized, and misappropriated, storage capacity. Managing these disparate storage resources became increasingly difficult and costly for the university, says Livingston.
UUHSC’s IT division started looking for ways to consolidate its SAN resources, with the goal of eliminating stranded storage and centralizing storage services as much as possible. Virtualization surfaced as a means to this end.
“Virtualization has helped out a lot,” says Livingston. “It helped us basically get rid of stranded storage.” Capacity from the various SAN islands was virtualized, or consolidated, behind a Hitachi Data Systems TagmaStore Universal Storage Platform (USP) disk array, which meant that pooled storage was now available to the university’s 200 disparate applications as needed (see figure, below).
Livingston estimates a capacity utilization rate of 50% to 90% before implementing the TagmaStore USP, which provides array-based virtualization. By virtualizing the storage environment (and gaining access to stranded data), Livingston expects to be able to reclaim as much as 18TB of stranded capacity over the next three years and to lower management costs by more than $300,000 over the next three years (see “at a glance”).
In addition, Livingston says virtualization has enabled the university to implement a “tiered” storage environment. By better aligning data resources to SLA objectives (i.e., to ensure that data with high-performance demands consumed high-end disk capacity and less-important data, with less-demanding retrieval objectives, is stored on lower-end SATA disk, for example) and providing shared access to a single capacity pool, Living-ston says he will be able to further reduce operating costs.
“Now that we’re able to tier our storage, we’re able to buy the appropriate level of disk for the appropriate level of data,” says Livingston.
UUHSC expects to see a total savings of more than $1.5 million over the next three years and a return on investment (ROI) of about 127%. Livingston estimates the cost of the HDS virtualization implementation at about $705,000.
While the university standardizes on HDS for both hardware and software, it also plans to virtualize two existing non-HDS disk arrays in the next few months: a StorageTek FlexLine V2X2 and an IBM “Shark.” Hitachi’s TagmaStore USP platform also allows users to virtualize disk arrays from EMC (Clariion CX Series) as well as from Hewlett-Packard (StorageWorks EVA).
The university plans to implement HDS’s HiCommand management software to manage its heterogeneous storage resources, although it is currently relying on various platform-specific tools for management.
The UUHSC IT environment consists of about 500 servers (mainly Windows) and approximately 200 applications and 50TB of storage capacity.
AT A GLANCE
- Extended useful life of older storage devices, allowing capital expenditures to be spent on high-end storage capabilities
- Enabled tiered storage
- Reduced stranded capacity
- Will enable consistent management from a single platform, including administration of heterogeneous disk and fabric elements
- Decreased staff count by two
- Reduced overall capacity requirements; put off storage purchases of 18TB for three years
- Allowed three storage frames to be removed, which translated into hardware savings of $200,000
- Enabled tiered storage
- Increased application availability
- Total projected savings: approximately $1.6 million over three years
- Payback in 14 months
- ROI: 127%