Another nail in tape's coffin

There's a lot going for tape. Apart from the conditioned rooms it's advisable to keep them in, and the need to refresh them every couple of years, and to carefully manage the encryption keys that you used to protect the data on them....they're ideal. But seriously, there are a lot of benefits to using tape, mainly high capacity and low maintenance and energy costs.

Problem is, people are buying fewer of them. There are some new sales figures out from the Santa Clara Consulting Group which show a serious slide in popularity of enterprise backup tapes over the last quarter. Although sales of LTO5 tapes were up 22 percent, LTO4 and lower were all down, and smaller tapes such as DDS/DAT (remember those?) were down too. Overall, tape cartridge sales amounted to $220.07 million in the first quarter of 2011, but fell to $185.5 million in the second quarter -- that's over 15 percent down.

There's only one conclusion to be drawn: unless people need to back up huge chunks of data -- one LTO5 tape can store 1.5TB uncompressed -- they are opting for cheap capacious SATA disks. That means most enterprises are buying virtual tape libraries instead; Hitachi has just announced that it'll be shipping a 4TB disk with 1TB per platter real soon now. As well storing more, disks are more reliable too, being sealed up -- but you know all this stuff.

Although there are drawbacks to disks, energy consumption among them, there are ways around this by allowing the disks to spin down. After all, this is deep backup, not daily production data so an access time measured in minutes rather than milliseconds is not so critical. And that's still way faster than the time it takes to fetch a tape from a library, load it, and shuffle to the right place on the serial medium, assuming that the tape is available and not offline in an archive somewhere.

Meanwhile the very large enterprises who have invested serious amounts of money into tape libraries along with the accompanying hardware and software, and who are the main buyers of technology such as LTO5, will be wondering if disk isn't the way forward.

Our descendants will wonder why we stored data on so flaky a medium as tape -- much as we do today about cassettes (although reel-to-reel is another matter!). And they won't be able to read those tapes, any more than the corporations who created the data in the first place will unless they've kept working examples of every tape drive they have ever used, along with the associated drivers and software, and a computer that will still accept ancient SCSI cards.

So much for progress.