Las Vegas Review-Journal’s class-of-service storage management

Brings a publisher scalability, manageability and savings.

At Las Vegas Review-Journal, company growth had turned data storage into a huge management challenge. By addressing the problem using innovative class-of-service technology, IT has eased the storage burden while saving the company money - and netting it a 2006 Enterprise All-Star Award.

At Las Vegas Review-Journal, company growth - largely through acquisition - had turned data storage into a huge management challenge. By addressing the problem using innovative class-of-service technology, IT has eased the storage burden while saving the company money - and netting it a 2006 Enterprise All-Star Award.

Steve Olson, infrastructure manager at the Las Vegas-based publishing group, describes his quandary. The most business-critical editorial, advertising and accounting data was stored on a 4TB EMC Symmetrix DMX 8530 array and a 1TB EMC Celerra network-attached storage (NAS) array. Desktop data - files, e-mail and archived documents - was mostly stored on about 60 Macintosh, Solaris and Windows servers at 40 sites in nine states, but some of it resided on the high-end Symmetrix. Storing nonbusiness-critical desktop data on expensive primary storage didn't make sense; neither did buying more servers with direct-attached storage, which would have complicated the management problem, he says.

"We needed storage that didn't need to be as fast or as expensive for our Tier-2 data," Olson says. He wanted a system to which he could move the less business-critical data being stored on the Symmetrix as well as for the messaging, home directory and file data being stored on the servers.

A classy storage strategy

Olson found his answer in a storage-area network (SAN) and NAS combination. He could choose from a host of vendors, such as storage heavyweights EMC, IBM and Network Appliance. But start-up Pillar Data Systems grabbed his attention, especially when the company talked to him about CoS available with its Axiom 500 midrange storage system, he says.

"With class of service, we can prioritize servers and guarantee speed of access."

- Steve Olson, infrastructure manager, Las Vegas Review-Journal

With CoS, Olson would be able to prioritize a server's access to disk according to the business criticality of the application running on it. That means he would be able to use Pillar's system to assign business-critical Tier 1 production databases to the highest priority of I/O while giving less important applications, such as a messaging system, a lower priority, he says. Prioritizing data in this fashion would let him store the most important data on high-speed, expensive Fibre Channel disk and the less important data on inexpensive and slower Serial Advanced Technology Attachment disk drives - both available in the Axiom system.

"My biggest fear was that because we have a shared set of resources . . . an SQL server [query] was going to get stepped on by somebody running a personal media clip," he says. "With class of service, we can prioritize servers and guarantee speed of access for the servers" that need it.

Olson points to another reason he favored Pillar - its disk-stroking software, which lets him determine where on the disk to store data and how quickly his 1,500 users can retrieve it (prioritized I/O).

Olson installed the $200,000 Axiom 500 system in April, and says he's already seeing cost savings. The 20TB Axiom cost the same as his annual maintenance and license fees for the EMC Symmetrix DMX 8530, for example. To scale the Symmetrix, he would have had to buy a new box. But expanding the Axiom system to a high capacity of 384TB simply means adding disk drives.

Storage simplicity

Using Pillar software, Olson has migrated data from as many as 45 Windows, Solaris and Macintosh servers to the Axiom system, giving him the benefit of managing a single storage environment, he says. And, because he has a single pool of information rather than 45 separate storage repositories to look at, he can properly analyze capacity and use and determine when he needs to add disk drives.

"Since data is now all centralized, it will be easier to manage and monitor," Olson says. "I also have a better justification for scaling growth, as well as a centralized environment for backing up the network."

For backups, Axiom runs software that automatically migrates data from one volume to another. The AxiomOne Volume Replicator software, which is based on storage-management technology from Kazeon Systems, copies data on the Fibre Channel SAN. The AxiomOne File Replicator software, available through a deal with remote data-protection vendor Signiant, copies data on the NAS side.

With the Axiom system, Olson also plans to replicate data to unlike machines for disaster recovery, a capability he would find difficult, if not impossible, to do with an EMC combination. "EMC's [Symmetrix Remote Data Facility] environment is very expensive. It requires a whole bunch of McData channel extenders and Brocade switches," he says.

Olson says he is testing Pillar's replication capabilities with the hope of making remote disaster recovery a possibility.

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2006 IDG Communications, Inc.

IT Salary Survey: The results are in