To most corporate systems administrators, disaster can be spelled many ways: fire, flood, hurricane, tsunami, espionage, and human error, for instance.
To prepare for such events, managers are turning to a variety of disaster-recovery solutions, ranging from contracting a hot site to the real-time transmission of data to a remote site.
In a typical disaster-recovery scenario, systems operations move to a contracted hot site using data restored from backup tapes. But what if the data on those tapes is a week or even a day old? Newer, crucial information may be lost.
Information managers with time-sensitive data are finding that new solutions, such as electronic vaulting and server mirroring, can provide the missing link in a disaster-recovery situation.
At Transamerica Life Companies, in Los Angeles, for example, Rimage Corp.’s Televaulting service is used to back up changed and new files from Transamerica Life’s 12G bytes of primary storage.
“What we do affects the livelihoods of a number of people, so it’s critical that we’re always on-line,” said May Nelson, LAN administrator for the firm’s Structured Settlements division, which handles annuity payments.
Rimage’s Televaulting allows Transamerica Life to transmit data via an ISDN line to one of Rimage’s five regional storage stations in New York, Atlanta, Minneapolis, Irvine, Calif., and Fremont, Calif. Equipment needs at the client end include a dedicated PC running a Rimage utility that automatically scans files for viruses, and then compresses and encrypts them prior to transmission.
Files are first transmitted to hard disks, and then copied onto rewritable optical disks.
“What we build for our clients is an incremental mirror. As the data comes in, it overwrites what was written the day before. We retain the last copy of all the files that were sent to us,” said Keith Stuessi, general manager for Rimage’s Minneapolis facility. “If users require a small restore or a large restore, the data can quickly be pulled off the optical jukebox and downloaded.”
As part of its service, Rimage also provides its customers with a monthly copy of their data on CD ROM.
Pricing on Rimage Televaulting is 50 cents per compressed megabyte. For a 100-user network that backs up every day, the cost is about $500 per month, according to Stuessi.
Vaulting doesn’t supplant backup
Electronic vaulting, however, has not replaced on-site tape backups for Transamerica Life. Palindrome Corp.’s Network Archivist software continues to provide daily backups onto a DAT autoloader, and tapes are taken off-site once a month, said Nelson.
“The Palindrome software tells you which tapes to take off-site,” she said. These tapes are stored in a fireproof safe maintained by Arcus Data Security Inc., in City of Industry, Calif.
Transamerica Life also maintains a contract with Comdisco Inc. for a hot site, in case a disaster prompts relocating the systems operation. This contract must be continually updated, however, to accommodate CD ROM and other new equipment.
“Right now, our contract doesn’t jibe with what we have,” said Nelson.
Most disasters are caused by human error, and good backups are often crucial. Televaulting services can also provide recovery from these more common disasters, said Stuessi.
Users can recover from the CD ROM or request that files be transmitted back to them.
Even the top hot-site suppliers are moving toward providing similar televaulting services. Comdisco offers Continuous Available Services, which duplicates the transactions of a customer’s computer equipment. In addition, later this month Sungard Planning Solutions Inc., of Wayne, Pa., will announce a televaulting program, said a company spokesman.
Every mirror tells the story
Another new solution to protecting data in the event of a disaster entered beta testing this month. Called Off-SiteServer, MiraLink Corp.’s $19,950 product allows users to establish a mirrored server in another part of the country.
Connected via a T-1 or E-1 line, the second server mirrors the first, recording all reads, writes, and bindery modifications. Everything is duplicated at the remote site.
Off-SiteServer includes two rack-mountable units with 500M-byte drives, Vinca Corp.’s StandbyServer interface cards, and mirroring software.
“This is a more reasonable solution than setting up a hot site,” said Keith Mixon, network technologist for GE Capital’s financial-services group, in Atlanta. “Since I run a NetWare shop, [OffSiteServer] appeals to me a lot.”
Mixon set up a second 4G-byte server at the firm’s mainframe data center, about 10 miles away from the main office. “If something happens, we can just go to the data center and set up shop,” he said.
Prior to this, GE Capital simply stored its weekly backup tapes at its corporate data center. “The real-time support we gain with Off-SiteServer is critical,” said Mixon.
Test, retest, and test again
Thorough testing of a disaster-recovery program is crucial to determining its effectiveness.
Apollo Travel Services, a technical-support center for nearly 20,000 travel agencies, routinely tests its disaster-recovery capabilities by closing down its central operations and moving to its Comdisco hot site.
“In a disaster-recovery scenario, we are primarily interested in restoring data from our file servers,” said Lynn Overstreet, business continuity coordinator for Apollo Travel Services, of Atlanta.
The Comdisco hot site is able to replicate Apollo’s 21 file servers, WAN, and connection to a Denver mainframe. Data is restored from weekly tape backups kept by Comdisco.
In case of an emergency, executable software and stable data are also kept at the Comdisco hot site.
“We do all the hardware and software technical support for close to 20,000 locations. There is no other center that could back us up,” said Overstreet. “Comdisco was able to replicate our operating environment very closely.”