A storage manager was showing me the latest and greatest in his data center a few months ago, and we got to his tape library. Scorn and frustration crept into his voice as he described his recurring problems with drives in the library. Periodically, tapes would jam during backups. Calls to the library vendor resulted in a service tech coming in, yanking out the offending drive--tape still stuck in it--and replacing it with a new drive and cartridge.
After several episodes of this, the storage manager was older and no wiser as to the cause of his problems. Were the drives used by the library manufacturer a bad batch? A defective design? Poorly manufactured? Were the tapes bad? There was no way to know.
We hear a lot of folklore about tapes and drives from storage managers. Certain makes are rumored to be better than others. Some tapes are sold with longer guarantees, yet it's not very clear how different they are from the tapes sold with shorter guarantees. Back in the day, guys in high school used to sit around arguing whether Ford stood for "Figure On Repairs Daily" or GM really meant "General Maintenance." At least those guys knew what they thought--the question of what causes tape failures seems to produce more shoulder-shrugs than rants. That's why we've decided that less talk and more testing was needed to shed some light on this vexing issue.
Here you'll see our initial effort. We've partnered with Mountain Engineering II. This first effort is provocative and admittedly incomplete, but we have discovered that there are causes for concern about the physical condition of LTO tapes, and that there are differences between any two tapes. Our sample was far too small to make that conclusive. Still, we felt that you deserve to know what we've learned already and that we will be pushing these tests farther, looking at broader samples. We'll also be looking at SDLT tapes, which should have similar reliability characteristics to LTO.
As we head toward terabyte tapes, you need to know more than ever how likely it is that the tape you buy will not fail in use. That's even more true at the increasing number of shops where tape plays a crucial role as a long-term archive tool. What good does it do to spend millions of dollars on software and systems development if data is lost because a $100 cartridge was made to manufacturing tolerances that were too loose, or was made outside of those tolerances but got on the market due to poor quality control?
I'm reasonably sure we'll hear unpleasant noises from the tape vendor community on this. Our goal here is not to indict, but to shine a light. Clearly, few customers of tape media vendors either feel good or knowledgeable about the quality of tape. Tape vendors, let me say this in advance: Before you yell that we're crying "Fire," check out why there's so much smoke.
We want to hear from you on what your largest problems and questions are with tape. Do you find that cartridges from some vendors work better than others? Do you have the tape drive from hell--that drive slot in your library that's always breaking down? Please let us know at firstname.lastname@example.org.