This is a copy/ paste from an update I gave in an old thread ... excuse me for not typeing it out again, but I'm in a hurry and have to dash out ...
With regards to StorSentry, not sure who markets this at the moment, it ws sold through Imation but I think that has changed.
****************
I did an approx year long investigation on this same subject and the tape manufacturer 'Imation' was also involved. I have also spent time at the Imation lab in Germany, speaking with some of their tape experts.
Although the manufacturers quote lifetime in terms on number of mounts we concluded it was not an accurate way to meansure tape condition. When I say we, this included one of Imations tape experts, not someone who knows a bit about tapes, someone who knows a lot ...
It is not so much the number of mounts, it is how the tape is treated when it is used, how it is stored, how it is transported.
Put another way, if you 'shoe shine' a brand new tape (that is, run it below the streaming speed) you could, with the right software, pick up the damage this has done is say a couple of weeks. Further, if you shoe shine a tape, the internal tape pack as it is known, becomes irregular (think of unrolling a roll of paper kitchen towel and then rolling it back up again - the two ends are no longer flat) - same with tape, some of the wraps of tape protrude from the pack. When the tape is moved, the vibration can cause the edges of the tape to become creased if then knock against the inside of the casing - known as edge damage. Under just the right conditions, even the vibration cause by the tape being moved by the robot can cause this. (I know this as we had edge damage on tapes that had never been removed from the library, so that was the only possible explanation).
Put a tape into a bad dive, it could cause tape damage instantly, or accelerate the wear on the tape - nothing to do with the number of mounts.
More usful is the total end-to-end writes, but as a tape is probably not filled every time it is written, this is pretty impossible to measure.
It is simply, an almost impossible thing to measure, without specialist software, such as StorSentry.
Studies using StorSentry showed that companies, that replaced all thir tapes after 3 years (just to be safe) in fact would only have to have replaced around 4 - 5 % of their tapes, that is, only this percntage showed issues in terms of error rate.
For the record, I'm no involved with StorSentry, but have used it and it is very ggod. It monitors tapes/ drives at a 'low' level, and alerts when the tape/ drives hit a preset (but adjustable) limit. That way they can be swapped out before failure.
Sofftware like this (there are others) is the only way to determine the true state.
You could monitor the /usr/openv/netbackup/db/media/errors file for any media/ drives that apppear frequently.
I wrote tperr to assist with this:
https://www-secure.symantec.com/connect/downloads/...
It uses statistics, so is more accurate the bigger the errors file, and has proved, on average, to be fairly accurate. Unfortunately, I've only tested it on Solaris.
Another option:
http://www.imation.com/en-US/Scalable-Storage/Scal...
This uses the data held on the internal LTO chip to ideentify bad drives/ media. The first version of this involved Imation coming to site and manually scanning the tapes and running analysis on the results. In my case they pulled out 15 tapes from about a 1000 that had issues, so I've seen it and can confirm it works.
If your tapes are treated perfectly, never run below streaming speed, kept at the exact correct temperature and humidity, in a 100% clean environment and are never transported, then yes, I would agree the number of mounts would be more meaning full.
In my experience, very few companies can guarantee this, and hence the usefulness of no of mounts is reduced.
Oh, I just remembered, StorSentry also reports on drives that are 'show-shining' tapes ...
http://www.quotium.com/prod/storageManagement.php
Some of the screen shots are in French, the software is written by Hi-Stor wen is a French company.
