The Role for Tape

10:30 AM -- As I wrote in a prior entry, tape's biggest challenge is that it needs an advocate in the manufacturer community, someone that will unabashedly promote the use of tape. Beyond this, tape also needs to define its role. The tape drive manufacturers will claim that the ideal role for tape is as a deep archive medium.

I think tape manufacturers are falling into a trap in the archive market. Certainly there are some ideal aspects to using tape for archiving; it is cheap, deep, and, when not in the drive, it is green. I think the biggest problems for tape are lack of trust and lack of search.

Trust is a confidence issue. Tape has to overcome the war stories that IT professionals like to tell. We all have our favorite "the day tape let me down" story. Belief in increased reliability is one of the major drivers of the disk-to-disk backup market. Somehow tape manufacturers have to prove that current tape technology is reliable and something that can be counted on.

Lack of search is a bigger issue and is more than one of perception. Depending on what you are using to archive to tape -- either your backup application or a specific archive application -- there will be some level of search built in. The problem with backup databases is always size, and most users don't keep the full detail for an extended period of time, certainly not the decades that archive may require. Archive applications may have a simpler time, but they are not also tracking hundreds of incremental and full backup jobs.

The key challenge is neither backup nor archive apps are designed to build a long-term, context-level search index. After a few years, most people are going to forget the names of the files they store and are going to need to reference data by a contract number, vendor name, or some piece of information in the file. Systems from vendors like Index Engines Inc. can get around this by providing search and extract of tape, but this still leaves the requirement of having the backup or archive application around that wrote the data. Unlike disk archive systems that typically store data in an NFS- or CIFS-mounted file system, most of the tape-based applications store data in a proprietary format. Anyone change his or her backup application in the last five years?Another issue is the verification of tape over the long haul. Unlike disk archives that have the ability to perform routine data continuity checks, most tape systems do not have an automated verification or migration function; one would be needed to sustain a long-term archive.

My opinion is that tape may actually have a better role -- a role it is already serving -- as a medium-term backup target. A sample process could be disk-to-disk backup holding 30 to 90 days worth of backups, tape holding 90 days to two or three years and then disk archive holding everything else for the next 80 years.

The big problem with disk or tape archives is actually reading the data recovered 50 years from now. I'll have some thoughts on how to do that on another day.

George Crump is founder of Storage Switzerland , which provides strategic consulting and analysis to storage users, suppliers, and integrators. Prior to Storage Switzerland, he was CTO at one of the nation's largest integrators.