Tape drives aren't immortal, but estimating how likely a drive is to fail in your application is more complicated than reading the manufacturer's MTBF numbers off the spec sheet. A back-of- the-envelope number for estimating the probability of failure is calculated from the data transfer rate, the annual failure rate (which is easy to estimate) and the actual duty cycle of the drive in your application.
You can get a first estimate of the failure rate by dividing the drive's MTBF by 8,600, the number of hours in a year (8,760 if you want to be precise). However unless you consider the drive's duty cycle this estimate can be off by a factor of two or more.
MTBF for tape drives is calculated at a given duty cycle, that is the percentage of time the drive is actually reading, writing or verifying data. The duty cycle used in the calculations is critical, because if your actual duty cycle is significantly different from the manufacturer's assumed duty cycle, the MTBF will vary as well, sometimes non-lineally. This is especially true if your actual duty cycle is higher than the one the manufacturer used.
You can estimate the actual duty cycle by dividing the amount of data to be backed up by the data transfer rate (in MB/sec) and converting the result into percentage of 24 hours. ([Data amount/Transfer Rate]/[3600/24]*100. If you had 100 GB of data to back up, and a 10 MB/sec. data-transfer rate, this would give you a duty cycle of 11.57%). Actual duty cycles vary widely between enterprises, or even among servers within enterprises. Typical figures range from 10 to 60 percent or more. Similarly duty cycles used by manufacturers in their calculations can vary from 10 percent to 100 percent.
Because duty cycle is critical in understanding MTBF numbers, most manufacturers include the duty cycle number in their product literature, although you may have to dig a little to find it.
A paper explaining the method and its applications is available at: www.dynamic-imaging.com/products_services/ docs/ratingtd.doc.
Rick Cook has been writing about mass storage since the days when the term meant an 80K floppy disk. The computers he learned on used ferrite cores and magnetic drums. For the last twenty years he has been a freelance writer specializing in storage and other computer issues.