I was somewhat surprised by Quantum's introduction last month of a new deduplicating Virtual Tape Library (VTL). Ever since backup applications started universally supporting backup-to-disk, I've wondered how long the virtual tape library market would continue to be viable. Treating disk like tape adds complexity to the data path, and VTL vendors make a significant markup on the software and integration that turns a server with a bunch of disk drives into a VTL. Are storage admins really so conservative that they won't redefine a few backup jobs to save $20-100,000?
I must admit, I was excited when I first saw Quantum's DX-30 back in 2002. With the new and wondrous virtual tape library (VTL), I could use the backup software and processes I'd spent many long nights getting to work most of the time without worrying about tape library jams. Since I was backing up to disk, I'd never again have tape drive shoe-shining, turning a one-hour backup into an eight-hour job.
By the time I ran a VTL bake-off for Network Computing in 2005, backup software had started supporting disk backup, and I figured people would buy VTLs as quick-fixes for their broken backup architectures for about five more years. During that time, we'd see a transition to backup to disk. Its five years later, and I still think the VTL market has five years to live. After all treating disk like tape means living with all the limitations of streaming tape systems especially treating each tape as an atomic entity. Tapes can't be deleted, just overwritten or discarded. Even worse, you can't erase some of the data on a tape.
Imagine you hire your competitor's star salesman, or Mark Hurd, and he takes his contacts with him. A judge orders you to dispose of all the stolen data. How do you delete it from backup tapes? I once got a call from a company whose web server was taken over by hackers that used it to store child porn. After much consultation with lawyers on both sides, we ended up restoring each tape, deleting the offensive material and spooling it off to tape again. Lots of billable hours, but otherwise a waste.
Backup software vendors also have to bear at least some responsibility as leading products like NetWorker, NetBackup and Backup Exec are still organized around the concept of atomic backup containers. For the most part, they just substituted container files for tape cartridges, still treating them like tapes, managing them in media pools and enforcing retention on a container by container basis. Some even leave containers that have exceeded their retention period on disk to be overwritten like tapes rather than deleting them like the files they are to return space to a free pool.