"Tape is getting attacked from both ends," he says. "It is getting squeezed between optical technology and hard disk-based [nearline storage], and the cost advantage of tape starts to become unimportant; over longer periods of time, the natural advantage of optical-longevity, durability and absolute authenticity-come into play."

Your best bet for archiving?

Can one company single-handedly make optical relevant to large numbers of customers? Not likely-particularly in Australia, where optical jukeboxes remain a niche technology and most customers use tape for backup simply because it has been the only viable option.

Lachlan MacDonald, who for 15 years has served as managing director of Gosford-based storage integrator AustSTOR Data Storage, believes the 50-year life of UDO and high capacity of multiple-UDO jukeboxes will convince many customers to give optical a second look. To hammer home that point, AustSTOR recently began offering a 1TB UDO jukebox solution, built around a 33-drive, 24-slot Plasmon G-24 jukebox and QStar hierarchical storage management software, as an integrated archiving solution for $18,000.

MacDonald concedes tape is strong in many markets, but says more customers need to become aware of the differences between backup-in which a huge amount of data must be dumped to storage as quickly as possible-and archiving, which requires large volumes of storage to be kept online and provided to users.

"Tape as an archival technology is still under the microscope," says MacDonald. "[Despite the recent introduction of WORM-styled tape] it still doesn't offer true write-once protection. Tape is searching for an alternative market to backup. While in certain instances it will meet compliance requirements, many companies stipulate optical for archiving to get true WORM."

Since genuine archival applications are relatively few and far between, Australia's market for optical jukeboxes is tiny -so small that market watcher IDC hasn't yet bothered to study it. Here, HP is just about the only vendor directly supplying and supporting optical jukeboxes, although resellers like AustSTOR give Plasmon a local presence as well (as an interesting aside, HP actually buys its UDO jukebox drives from Plasmon).

Locally, HP also runs a big business selling tape jukeboxes into its large base of corporate customers -a duality that would suggest HP is less partial toone technology over the other.

So, which technology does HP see as better? Ian Selway, marketing manager for ILM with HP, believes many customers are buying tape when they might actually have been better served by an optical solution.

"Tape is designed as a disaster recovery medium," he says. "I think that for years we've been potentially misusing tape as an archiving medium. It is not good as an archiving medium; restore times for tape are such that you don't really want to be doing to a tape you wrote a year and a half ago to get a file from it."

Building a long-term strategy

Although tape wins in absolute capacity and cost per capacity, other specifications suggest optical can still hold its own. Today's tapes last 15 to 30 years, while UDO is estimated at 50 years or more-but in neither case should this be a deciding factor. After all, despite manufacturers' claims of storage permanence, customers are likely to have to regularly migrate their data to new formats well before the media degrades, if only to maintain backwards compatibility.

"What appears to be a satisfactory medium for a very long time may not be so," warns Graham Penn, associate vice president for storage with IDC Australia. "Even tape that was current 10 years ago is barely readable today. No matter what technology you use, if you're in a very long-term cycle you have to build the technology refresh cycles into your costs and plans."

Comparison of technical specifications supports the idea that optical and tape are for different purposes. UDO's 8MB/sec transfer rate is lower than the estimated 70MB/sec of LTO-3 drives, but finding a specific piece of information on a tape can take minutes as hundreds of metres of tape wind. UDO, by contrast, offers a 5-second media load time and 35 millisecond seek time. This makes it less effective than tape as a straight backup medium, but random-access capabilities make it eminently more suitable for recovering specific pieces of information that may be scattered from one side of the disk to another.The challenge for customers is to decide whether having this rapid access is worth the extra complexity of adding an optical jukebox into the enterprise storage mix.

Increasingly popular mid-tier disk, comprised of inexpensive SATA hard drives, must also be considered because it is easily integrated and provides even faster access to archived information. However, SATA drives cannot offer the permanence or easy transportability of an optical disk, and their lack of WORM capabilities makes them suitable less as long-term compliance tools than as a short-term buffer between primary disk and tape or optical.

In an ideal world, customers might backup their data to tape but archive key pieces of information-emails, financial figures, or the last year's worth of customer documents for easy retrieval, for example. However, Joan Tunstall, marketing manager with StorageTek (currently the subject of a $US4.1b takeover bid by Sun Microsystems), isn't losing sleep over optical just yet.

"People are typically using optical because of the WORM capability, but you've got to look at where the industry is going," she says. "Fully 99.9 percent of archived data is on tape, and we don't see optical coming into the picture in the future any more than it is today.

Optical will play in smaller shops, but I don't believe anyone looking at enterprise class archiving would consider it."

Optical advocates aren't pinning all their hopes on UDO, however: Iomega researcher Fred Thomas recently unveiled a blue-laser technique that uses subwavelength features within data pits that should offer 250GB per disc in the first generation and could eventually-theoretically, at least-provide more than 300 times the capacity of UDO.

Then there's InPhase, the Bell Labs spinoff that is developing a completely new archival format based on holographic storage. In this model, which was demonstrated in prototype form in April and is due on the market next year, interference from collisions of two lasers is recorded onto a light-sensitive disc as 'pages' of more than 1.3 million bits each.

Pages are layered into 'books'-InPhase has already demonstrated 2500-page books-in a design that is expected to deliver 800GB discs by 2008 and1.6TB by 2010. Remember, that's uncompressed data. By comparison, LTO-4 is expected to reach 800GB of uncompressed data per tape around 2007 and the subsequent LTO-5 drives will max out at 1.6TB of uncompressed data (assuming, of course, that makers can continue shrinking already microscopically thin tape). Around 2009, fifth-generation LTO-5 tape drives will store 1.6TB of uncompressed data and transfer 180MB/sec uncompressed data-not much better than the 1.6TB holographic discs expected to be moving data at 120MB/sec by then.

"We are really targeting this competitively against tape," says InPhase vice president of marketing Liz Murphy, who believes holographic storage will find a natural home in broadcast, medical and other verticals requiring massive amounts of online data.

In a 638-slot jukebox like Plasmon's top-end G638, holographic storage would deliver 1020TB of online optical storage. That's enough to bridge the gap with tape, letting customers finally choose optical or tape based on capabilities such as fast random access. If tape vendors aren't losing sleep over optical now, perhaps they should set their alarm clocks for 2009.

Related Article:

IBM and HP ready to unleash LTO3