Media Wars

The need to back up and store enterprise data is undeniable. How to do it is the question. Batch processing or continuous data protection? Point in time imaging? RAID disk arrays? Tape? Disk? A combination of the two? Optical media? Memory sticks? A length of fencing wire and some sticky tape?

Even Garry Barker, chairman of the board of directors of the Australia/New Zealand branch of the Storage Networking Industry Association, admits that "there does seem to be a battle between tape and disk and tape plus disk".

Despite the enthusiasm with which suppliers espouse the advantages of their various and competing technologies, Barker says there is no generic answer to the question of disk or tape. "The industry hasn't come up with an answer . . . a battle is raging."

Tape has been the traditional medium of choice. It has been around for ages and has a good record for longevity and reliability if handled correctly and managed properly. It has its downsides, most notably the gap between starting a backup and when it is completed, leaving what is referred to as a "backup window" in your data. This is why most tape backups are done overnight or, even better, over the weekend. But it still remains one of the key issues picked on by suppliers of alternative media.

The other issue in enterprise backups is the sheer size to be managed.

Not only do current regulations insist that more data than ever is stored, but organizations themselves are increasingly realizing the potential value of information, even if they do not know exactly what it might be. Better to store it, then, just in case. Changes in technology - such as the improving resolution of visual images - mean files are increasing in size, requiring more storage space. In addition there is the general increase in data being generated by you, them and everyone else. Finally, there is just the shear bloody-mindedness of people who wish to store everything for a rainy day, as if data storage was the jammed backyard shed of an organization.

Victor Koshka, South Pacific regional manager for BakBone Software, says that volume is the biggest danger facing enterprises. "Some day it's going to implode somewhere. We're just chasing our tails."

Take Mark Adams's dilemma, for example. Adams is network and systems administrator for Film Australia, the government-owned organization that has been producing documentaries and educational programs since World War II. Film Australia has a program to digitize all of its old stills and films, a project that should take about 12 to 18 months to complete. Film Australia's library has 150,000 images and 5000 films, covering a century of Australian history, so digitizing this is a massive undertaking, but Adams says it is nothing compared to the day-to-day data issues he faces.

"It used to take us 24 hours to do a backup. The old tapes were not big enough." He recently installed a new tape backup system (LTO3), which allows him to cut the backup period down to a more manageable five to six hours.

He uses tape for long-term storage, and disk for short- and medium-term (the biggest holds 3TB), and down the line is looking at disk-to-disk backup, as well as disk-to-disk-to-tape. He now has faster backup and larger capacity, he says, adding that "It's not bad that disk prices are coming down".

Richard Giddey, A/NZ country manager for Exabyte, thinks storing corporate data is manageable, but "storing personal data on company systems is out of control . . . and probably wrong anyway". Which means that whether you go for disk or tape (D2T), D2D or D2D2T, the real element in enterprise backups that is of concern to many is the human one.

"There is a lack of people skilled in doing backups," Koshka says, and others agree.

Craig Tamlin, A/NZ country manager for Quantum, relates the story of a multinational's office in Papua New Guinea. "A field agent was put in charge of backups because he was the most technical person available. But he didn't understand the implications of what he was doing, and soon got sick of going to each machine in different offices and changing the tapes. He noticed that when each tape was full, the machine would spit it out. So he decided to put masking tape over the cassette feeder. Each time the tape was popped out, it just went straight back in again, to be written over again and again. A perfect solution, he thought, until he managed to save (and resave) a virus, so that when it came time to restore there was nothing but corrupted data."

Giddey endorses this concern: "90 percent of tape failure is because of data not being written to it in the first place." And Barker suggests that most faults in storage are found in people or software, not the hardware.

At least enterprises are actually backing up. "They'd be silly not to," Koshka says, but he quotes a recent survey that indicated 50 to 60 percent of SMEs are not backing up at all.

Of course, backing up is only part of the story. There is also recovery, which, in the long run, is far more important. A recovery can be the real problem or, as one vendor put it in describing incremental backups, "you get a faster backup, but it's a bugger to restore!"

Barker sums it up by saying that it is important to look at the totality of the problem. "There are many different objectives. Backup and retrieval are not the same as continuous availability. Don't lock yourself into one technology solution."

And for those who think storage is more work than you really want . . . "It's fraught with failure and false starts but it's getting better. It's getting more exciting and more fun all the time," Barker says.

And it is certainly getting more important.