During the debate over archives for regulatory compliance, one member of the storage hierarchy has often been missed out: optical disk. Yet according to its proponents, optical is the natural choice for archiving.
"My long term archive has to be very stable, I don't want to migrate data or pay a lot for maintenance, and I want properties that aid with compliance," says Steve Tongish, EMEA marketing director at Plasmon Data, which recently introduced a new optical format called UDO, using blue lasers to achieve 30GB per disk.
"70 to 80 per cent of the media we sell today is write-once because we play in a market where data is inactive or static," he adds. "Maybe the role of magnetic disk is expanding, but when data goes inactive and you want it on low-cost storage, optical is still the way to go. Optical is long-term - 10 or 25 years - so it still fits very well."
Tongish adds that as companies implement archiving for the sake of compliance, they should not forget that whatever they buy must still add value to the business. Cynics might think it is more about adding cost due to government regulations.
"You don't invest in compliance - you comply," he says. "Don't ignore the regulations, but you have to assess, manage and deal with them in a very efficient way so they don't become a burden. You need to get competitive advantage beyond just compliance.
"We talk about three drivers - one is compliance, another is protecting yourself against litigation, and the third is business benefit or value. It will probably settle into a more pragmatic role, you will have compliance officers in the organisations but it won't be the overriding issue. I'm not sure it's happening yet, but business value will re-emerge as the leader."
The business value in archiving lies in keeping static data available online at a low cost, he says:. "Often there's a misunderstanding about the difference between backup and archiving. Daily backups should not include static data - you need an archive strategy, not a longer backup window."
Backups therefore go to hard disk for short term retrieval and for staging to long term tape backup, while static data which needs to be accessible over the medium and long term is archived, preferably he thinks, to an optical library.
Why? Why not stick with tape and save the bother of managing an additional storage format? Tongish plays the access speed card.
"For backup, tape always wins because it's straight cost per GB," Tongish says. "But in archiving the data has to stay available. You might have an archival cycle with users hitting the database; for example Deutsche Bank uses optical and has hundreds of requests per day against its archive.
"They're random requests too, so they need quick access to an object, then they're done with it. An optical library is very fast, whereas a tape library would require a much higher drive-to-tape ratio for the same performance. Also, tapes need wear management and replacement, and they need a climate-controlled environment or else they suffer.
"The underlying message from Basle2 and so on is, you have a lot of data to store and provide access to. An email archive has to be static. It has lots of small objects that could be spread over three years for a single (overall) transaction.
"If you want to use spinning magnetic disk, I can demonstrate it'll cost three or four times as much. Disk is cheap but once combined, into EMC's highly redundant architecture, say, it's very expensive."
So tape is slow and wears out, disk is too expensive, leaving optical in center stage. Tongish ices this optical cake by pointing out that, despite the arrival of UDO, Plasmon will not abandon its older products. By comparison, data stored on magnetic disks will need to be migrated to new hardware every few years.
"That's the difference between the hard disk vendors and us - we take a different approach to support," he says. "We still have customers using old 12-inch optical, for example. We have to support those, and manufacture and sell them, because they're long-term storage.
"Optical reduces the frequency of migration to new archiving hardware, and that's what makes its TCO lower - you want it to last 10 or more years, compared to two or three years for a hard disk. Plus, hard disks can't be stored long-term as they suffer from sticking.
(That’s a new one on me. I Google searched on ‘hard drive’ and ‘sticking problem’ and came up with a reference.)
"For example, L'Oreal uses optical to retain records on cosmetics approvals. They took some disks out and stored them, 14 years later they were challenged in court on a product name and had to prove precedence. They cleaned up a drive and were able to find purchasing records that referred to their product name and won the case.
He warns that, in order to get this sort of media lifespan, you do need to buy the best media: "There is a remarkable difference in quality between CD and DVD media manufacturers. most are very consumer-oriented - CDs can last longer if you don't use cheap aluminium underlays, say.
"For our DVD libraries we insist customers either buy our higher quality media or spend the extra with other supplies, as you get what you pay for. We have lifecycle-tested UDO and are happy with 50 years, while it is $60 a disk, 30GB of your data is worth that."
Beware though that this 50-year lifecycle test will only be conclusively proved in 2054. (It’s probably best to store your Plasmon paperwork in a filing cabinet just in case the optical copy isn’t recoverable.)