RE: file position on tape

[Arie Shoshani <arie@math.lbl.gov>]

> In the long run, it may be that
> file order on tape may be one of the more effective optimizations.
> 
> Doug

That's correct, but it may be enough just to know the tape on which a 
file is is placed, if it is indeed true that HPSS will optimize read 
order of files out of the same tape.  

We have observed in previous runs that files are cached in an order 
different than the order of the PFTP requests.  We suspect (and Keith 
says it is supposed to be so) that this is because HPSS orders the 
PFTPs so that it minimizes going back and forth on the tape.

Arie.