Survey: Disk arrays for secondary data storage rising

Relatively inexpensive secondary disk storage is gaining a significant foothold in corporate data centers, according to a survey of more than 1,000 IT managers due to be released next month. But the survey also shows that a majority of data is still backed up and archived the traditional way: on tape devices.

Peripheral Concepts Inc., which released preliminary results of the survey this week, said about 50 percent of the respondents with disk storage capacities of more than 500TB indicated that they already use disk-based devices for secondary storage. That includes saving snapshot copies of data and staging information for archiving. Another 25 percent plan to start doing so within two years, the consulting firm said.

But the secondary disk storage typically involves less than one-fifth of a company’s total data, said Farid Neema, president and CEO of Santa Barbara, Calif.-based Peripheral Concepts. And only a small percentage of the data that’s backed up on disks doesn’t get moved to tape devices for archiving, Neema said.

“Tape remains by far the most popular medium and does not seem to want to go away,” he said.

Neema cited another potential roadblock to wider adoption of secondary disk storage: a “significant percentage” of the respondents indicated that reliability issues could prevent them from using low-cost Advanced Technology Attachment (ATA) disk drives.

Over the past 12 months, vendors such as EMC Corp., Network Appliance Inc. and Storage Technology Corp. have released ATA-based disk arrays for storage of near-line data, snapshot copies and information that will eventually be archived on tape devices.

Mike Lin, director of emerging technologies at the University of Southern California in Los Angeles, is testing StorageTek’s BladeStore virtual tape server, which acts as an intermediary between corporate systems and tape libraries.

Lin said the school currently backs up 18TB of research and administrative data to Sun Microsystems Inc. StorEdge 3510 disk arrays during an eight-hour window each night. It then archives the data on StorageTek 9940 tape drives.

A terabyte of Fibre Channel disk capacity for the StorEdge arrays costs US$65,000 to $90,000, depending on how it’s configured, Lin said. By comparison, he said a full 10TB BladeStore array with ATA drives costs only $40,000.

Lin didn’t voice any major reliability concerns about ATA technology, but he said it takes an hour longer – about eight hours altogether – to rebuild the data on failed ATA disks than on Fibre Channel ones.

Jamie Riis, CIO at BayView Financial Trading Group LP in Miami, is using a NetApp R100 NearStore disk array with 12TB of capacity to store images of documents. The array replaced an optical system that was slow and unwieldy, Riis said. BayView’s IT staff also uses the R100 as a virtual tape library for backing up database files prior to recording them to tape devices.