Geoff,

This is an issue that has been around since they implemented the DIRMC
option.  When I was working for an IBM business partner & I could attend
the internal seminars (in San Jose) I was told that the DIRMC option was
invented specifically for Novell clients since the programmer who wrote
that code had put all the subdirectory structure information into the
storage pool instead of the database (as is done by the other
clients).  Evidently if you add this directive to any client's
configuration you now store the data in the pool identified by DIRMC.  For
all clients, except Novell, the data is then stored in two places: the
database and the storage pool.

ADSM-L list participants that use this configuration tend to report better
performance for some processes (I believe migration and reclamation).

When I get some additional hardware that I can use for testing I'll take a
whack at evaluating this for myself.

Steffan

At 06:33 AM 3/23/2001 -0800, you wrote:

>I do not understand quite well what you try to accomplish. But anyhow,
>cached data on disk only remains until a new backup needs the
>space. In the
>tape you will have all the data from the different versions
>you are keeping.
>So it is normal that you have much more data on tape than on disk.
>If you want to maintain the data on disk forever you will need
>a very large
>disk storagepool.
>Regards
>Maria

I did find an answer to my own question after digging a bit. I do see that
cached files are kept on disk unless room is needed for newer files. This is
a 33gb pool that is only 0.6 percent full so for this data to not be here
when I queried for it was a bit confusing. Once I moved the data from tape
back to disk everything was fine again. I don't think my timing was off when
I created everything because the disk pool was in place for a few days
before I realized I did not update my scripts properly to send the data to
offsite tape and to tape that will remain in the library. Once I did that
the PCT utilized on the disk dropped by .2% so I was curious what was
happening.

This is just directory information so the amount of data may never come
close to filling the disk. In that case, the way I understand it, it will
"all" always be there. With the amount of nodes and size of my library I
can't collocate so I thought I would at least keep directory data cached on
a disk pool in case of a restore. I just might cut down on tape mounts and
restore time.

Please correct my assumptions if they are wrong. I'm still learning so any
help or criticism is welcome.

Geoff Gill
NT Systems Support Engineer
SAIC
Computer Systems Group
E-Mail:   gillg AT saic DOT com
Phone:  (858) 826-4062
Pager:   (888) 997-9614