On Mon, Mar 29, 2010 at 12:13, Ben Scott <[email protected]> wrote:
> On Mon, Mar 29, 2010 at 11:54 AM, Kurt Buff <[email protected]> wrote:
>> http://en.wikipedia.org/wiki/Linear_Tape-Open#Tape_durability
>>
>> Seems pretty durable to me - 4 years, if entire tape written once per
>> week. That seems unlikely.
>
>  Beware.  Understand what specifications are actually saying.  That
> particular scenario is not a directly applicable characteristic.
> Nobody accesses (reads or writes) every block on a tape once and only
> once before accessing it again.
>
>  When I researched this, any given spot on the tape was rated at X
> number of passes, individually.  If one spot on the tape is passed
> more often (for whatever reason), that still counts towards wear
> limits.  When you start getting errors on the tape, you retire it (at
> least, I hope you do), so it doesn't matter if 99% of the tape is good
> if 1% is degraded.  In particular, the beginning of the tape tends to
> get read/written a lot more often, because that's where the volume
> header is.
>
>  The table given in that Wikipedia article cites a link that 404's,
> so I can't even begin to determine what the manufacturer might have
> been talking about.
>
>  We sure as hell don't get four years out of our once-per-week LTO2
> tapes before they start generating unrecoverable data errors.
>
> -- Ben

Oh, yes. If you read/write to the first foot of tape,.it counts.

But, with our TSM system (with dual LTO3 drives in a 30 cartridge
robot), this was never an issue. Never had to discard any tapes

Now that we've switched to Ultrabac and a robot with a single LTO4
drive and 16 cartridges, we may have to pay a bit more attention to
it.

I suspect that doing D2D2T will help, though. We've got 3tb of disk on
the backup machine, and we're backing up roughly 2.5tb of data in our
fulls.

Kurt

~ Finally, powerful endpoint security that ISN'T a resource hog! ~
~ <http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/>  ~

Reply via email to