Problem solve Get help with specific problems with your technologies, process and projects.

Monitoring tape media duty cycle

Backup expert Ashley D'Costa offers advice on how to realistically monitor the duty cycle of your tape media.

Recently, we have gone to do a couple of restores and found that the tapes (SDLT 1) were bad. I have been trying to find information on tape life. I know that there are many variables, such as how well do the tapes stream, etc. I found one article saying that the tapes should be good for a 1,000,000 passes of the head. How you would calculate that? The only information I have is that I know we filled the library about two years ago. We have replaced some tapes that were frozen and others that seemed to always get errors, but we have a number of tapes that have been in since we received the ESL9595 library. I ran a report that shows me the number of mounts. How do I know when to remove tapes that may be getting unreliable?
This is a tricky question to answer. I've always found it hard to get a tape media manufacturer to specify realistically how reliable their media is in terms other than laboratory results (thus, the 1,000,000 passes number). It's the safest way for them to express the reliability of their product, although very difficult to get statistics on and, therefore, impractical to monitor. I will try to address the question by discussing how the issue is currently being handling in most IT shops and how to avoid errors in the first place.

Tape media duty cycle

With tape media, you typically don't know that a tape is bad until it fails -- at which point it's too late. As a result, the most typical way that this issue is addressed is to pull tape media out of circulation after a certain amount of time and assume it is unreliable even if it has not had any errors. This would be considered its useful duty cycle. I've found that linear tape technology (DLT-based/LTO-based) that is used daily typically gets pulled out of circulation after about a year. Helical scan technology is more prone to wear, and therefore, is pulled from circulation sooner -- generally after about three to six months (with the most extreme cases after only 10-12 passes).

Most backup/recovery products have the ability to enforce a tape duty cycle by attributing an expiry period to the tape media (this is different from the expiry period of the data on the tape that defines the data's retention). Because you state you have "frozen" tapes, I'm assuming you have Veritas NetBackup as your backup/recovery product since "frozen" is a NetBackup term. I know NetBackup does have an expiry date you can assign to your tapes, after which the tape becomes read-only until all data on it expires and then the tape is no longer used.

Archival life

A tape's reliability is also dictated by how long it has been sitting around -- its archival or shelf life. This doesn't come up as often with respect to tape reliability, mainly because of the long shelf life that most modern tapes can have. For example, depending on temperature and humidity, SDLT 1 has an advertised archival life of 15-30 years. Of course, this assumes no tape handling at all.

Tape handling

The best way to prolong your media's duty cycle is avoid having errors altogether. Errors are most times the result of damage to the physical medium within the tape cartridge rather than a result of a defect to the medium. This damage to the medium occurs most commonly because of incorrect tape handling procedures.

Tape media is typically handled because of offsite vaulting requirements. As a result, most tapes will be handled at least several times through the course of their duty cycle.

Modern tape technology (e.g., LTO-based, Magstar-based, DLT-based media) pack a significant amount of tracks on the medium -- far more than in the past. The margin for error is orders of magnitude smaller than 10 years ago. In the past, there were far fewer and much larger tracks. The margin for error was significantly larger, and therefore, tapes in the past could be handled more robustly.

Due to this legacy thinking, there is still a tendency to man-handle modern tape media the way it was handled in the past. For example, a common problem today is edge damage. If a tape is dropped, it is highly likely that the edges of the media could get crimped. Since, in most linear tape-based products, this is where the servo track information is kept (a track that allows the tape drive head to stay aligned with the tape) it is possible that media errors could result because the head can no longer "stay on track" so to speak.

If you have a high rate of tape errors, I recommend to reviewing your current tape handling procedures and then reviewing the best practices for handling the media from the tape manufacturer's Web site. DLT and SDLT media are manufactured by Quantum. Click on the following link for Quantum's very informative Care and Handling Guide (pdf).

Dig Deeper on Tape backup and tape libraries

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.