Lucas Mearian, Computerworld
23/07/2007 10:24:04
Fed up with slow tape backup systems and under pressure by regulators and auditors to keep data online and readily available, large and midsize businesses are making disk-to-disk backup technology a top priority in their data centers this year.
More than 75% of 150 large companies recently surveyed by TheInfoPro Inc., a New York-based independent research firm, said disk-to-disk backup technology is being used in their data centers; this compares to 67% who were implementing it a year ago.
Still, in the most recent survey, only about a third said they are using virtual tape libraries (VTL), a form of disk-to-disk backup that essentially uses disk arrays to mimic tape for server backup jobs.
One in four IT managers surveyed said they believe poor archiving practices are a key reason for unchecked data growth. The average company has 250TB of active storage space dedicated to archive-related content. Moreover, the survey found that archive capacity among companies is expected to grow by 52% by the end of this year.
Sean O'Mahoney, manager of client/server computing at Norton Healthcare, which has more than 2,000 physicians, rolled out three disk-to-disk backup systems over the past year. It helped the organization deal with a 50% year-over-year archival data growth rate that pushed the backup window to 20 hours a day.
Since connecting the health care firm's Picture Archiving and Communications System to an EMC Clariion Disk Library 710 array and Clariion Disk Library 4100 array, Norton's backup window has been more than halved -- to eight hours a day. O'Mahoney also plugged his company's financial systems into an EMC Centera disk array, which is a WORM permanent archive system.
In all, Norton Healthcare has 200TB of capacity dedicated to disk-to-disk backup, the vast majority of which is used for storing radiological images such as X-rays, which do not lend themselves easily to compression.
Besides shrinking its backup window with disk-to-disk technology, the organization also improved data-restore times by as much as 75%.
It used to take four days to retrieve data from tape because Norton's AIT-2 tape drives have a maximum throughput of 6MB/sec., compared to the Clariion disk array's 60MB/sec. rate. "The speed of the media is vital," O'Mahoney said.
TheInfoPro's survey, released in May, revealed that IT managers consider backup activities among the most time-consuming, adding that staffing remains flat and budgetary pressure to cut costs is at an all-time high.
A study by Gartner earlier this year predicted that by 2010, disk and not tape will be the primary medium for data recovery and that by 2011, the ability to take continuous snapshots of data will be an embedded function in backup and data replication software and will no longer be a separate feature.
Currently, only 17% of companies have deployed continuous data protection, according to TheInfoPro's survey.
"The need for high-performance online recovery of data, combined with the availability of low-cost disk arrays, has influenced enterprises and small and midsize businesses to adopt a disk-based approach for backup and recovery," Dave Russell, author of Gartner's disk-based storage report, wrote.
Tony Asaro, an analyst at the Enterprise Strategy Group Inc. in Milford, Mass., said the driving force behind disk-to-disk backup boom is a combination of massive data growth, which is expanding backup windows, and legal and auditing requirements that force firms to keep records online and accessible for longer periods.
"Everyone we know is doing some sort of disk-to-disk backup," Asaro said.
Most companies, however, are still relying heavily on tape backup for archive, keeping data on disk for 120 days or less, Asaro said.
O'Mahoney said he typically keeps data on his disk-to-disk backup systems for about three weeks. His company's auditors, however, recommend that he keep that data on disk for at least five weeks, something O'Mahoney is now planning to do.
O'Mahoney said the health care industry is struggling with how long to keep data on disk because "there's not a lot" of regulatory oversight. "As far as federal regulation of the health care industry from an IT perspective, we're way behind the public sector," he said.
While the Health Insurance Portability and Accountability Act addresses the security of patient data, it doesn't speak to data retention. Guidelines for data retention in the health care industry were simply carried over from the outgoing era of paper records, he said.
Some companies are even more aggressive about using disk-to-disk. About nine months ago, flooring manufacturer Shaw Industries Inc., a subsidiary of Berkshire Hathaway Inc.,deployed a NearStore VTL. The disk-based backup system from Network Appliance Inc. appears as a tape library to backup software.
In general, VTLs offload data far more quickly than tape drives do from primary disk arrays, which application servers use for shared external storage capacity. So users no longer have to quiesce application servers for backups.
Instead, customers can take repeated snapshots of data, offload them to the VTL and then at the end of the day -- or week or month -- store those snapshots on a tape library either on-site or off.
In Shaw's case, the VTL consolidates backups on the company's WAN, moving information from 24 remote sales offices around the country to the main data center in Dalton. The system currently has 10TB of capacity, but it will grow by another 20TB this year, according to Ben Worsham, Shaw Industries' infrastructure planning manager.
Worsham said he's been adding up to 40 servers a quarter, both in remote offices and in the main data center, and can no longer keep up with backups without disk-to-disk technology.
"The backup window runs pretty much 24 hours a day. We see adding capacity and servers to the VTL as a way to shrink that," Worsham said.
Like O'Mahoney, Worsham keeps data on his VTL less than 120 days before vaulting it to tape. He has yet to deploy disk-to-disk technology in his data center as a way to consolidate server backups, but he said it's one of the things at top of the list of new technology rollouts.
It's critical to get data backed up from branch offices to the main data center. "Primarily, our need to do this is from a [disaster recovery] perspective. You've got auditing issues," Worsham said.
Worsham said eventually the bulk of his company's backups will go first to his VTL and then be vaulted to a secondary disk array in an off-site location. But he's also convinced that long-term archive will always require some tape.
O'Mahoney said his disk-to-disk systems have shown his firm the potential to use other technologies, including storage resource management tools and storage virtualization. He said he hopes to someday roll out EMC's Invista switch-based virtualization software or IBM's SAN Volume Controller virtualization appliance.
"I think it is technology that has play in our environment, and I'll probably implement it over next couple of years." Being able to manage primary and secondary disk through a single interface allows for consolidation of assets, which will allow his systems administrators to manage greater amounts of capacity than they can currently.
According to TheInfoPro's survey of IT managers, almost 55% have deployed some type of storage resource management (SRM) software, and nearly 66% have deployed storage monitoring software.
SRM software manages physical and logical storage resources, including primary and secondary disk arrays, virtual devices and block and file-based data. Monitoring software provides a view into storage resources and how they're being used, but allows no control over those resources.
After seeing some proof of concepts that apply to his company's infrastructure, O'Mahoney also said his disk-to-disk systems will benefit somewhat from emerging de-duplication technology, which he hopes to roll out over the next year.
De-duplication ensures that only a single instance of structured or unstructured data is copied during backup. In many instances, the vendors claim, de-duplication can reduce capacity needs by as much as 80% on more costly disk-based subsystems.
De-duplication took the No. 1 spot on TheInfoPro's "heat index" of 25 technologies that companies are expecting to deploy more quickly than any other to meet their operating needs. "Forty percent of [Fortune 1,000] storage organizations have indicated that de-duplication is a key IT strategic enabler, given the difficulties of managing storage growth while maintaining flat staffing levels," the report stated.
Most companies said they were deploying EMC's Avamar de-duplication technology, followed by technology from Data Domain Inc. and Diligent Technologies Corp.
Previous - 1 2 3
[ Printer Friendly Entire Article ]
[ Other stories about IBM, ACT, Norton, SEC, Logical, Speed, EMC, Hathaway, Gartner, Network Appliance, HIS ]
No comments:
Post a Comment