Use both. Create a filepool with MaxScr=1000 over devclass with
MaxCap=2GB. Set a smaller diskpool in front of it with HighMig=10-30,
LowMig=0 and MaxSi=<something reasonable>.
It might be that the NFS repository works considerably better with files
under 32-bit limits.

Zlatko Krastev
IT Consultant






Mitch Sako <msako AT CADENCE DOT COM>
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
29.05.2003 03:35
Please respond to "ADSM: Dist Stor Manager"


        To:     ADSM-L AT VM.MARIST DOT EDU
        cc:
        Subject:        Disk Storage Pool Volume Size


I'm running a 5.1.6.5 Linux server with some NFS disk repository over
gigabit ethernet.  The NFS is pretty well tuned and stable and runs almost
as fast as local 10,000 RPM SCSI disk.  What I was wondering about is the
size of the disk repository data volumes.  I tried some 100GB files and
those were too slow.  I tried 10GB and that was a little better.  I'm
trying
2GB now and it seems even better.  Is there an optimal size for disk
storage
pool volume size?

The smaller the volume size the more management of the volume names, the
larger the volume size the less number of volumes, hence, easier
management.  I plan to have about 2TB of disk repository which means
1000x2GB files.  I'm just looking for some guidance for NFS-mounted
filesystems over gigabit ethernet.