From: Mike Knight, DCS-East, (314)234-5096, KNIGHTM at ISSCVM
Inet:                                       KNIGHTM AT VNET.IBM DOT COM
Subject: running time of auditdb; experience?

On 09/26/95, Helmut Richter wrote:

>Have you seen the effect that after some time there is a saturation from
>where onward the process is slower by orders of magnitude? If so, was there
>later a speedup again?
>
>What we saw is that the process was fairly quick on the first 13 mio
>entries and is now much slower, about 5 or 6 seconds per entry (compute
>yourself how long it will take for the remaining 37 mio entries).

We haven't tried an AUDITDB yet, but I was monitoring a restore this weekend
and had a similar symptom.  The restore was for 323,000 files.  The AIX server
saw the Solaris client session in SendW with increasing Wait Time as the
restore progressed.  Toward the end the wait time reached 21 minutes.  The
whole time, the client was running 100% CPU bound.  Supposedly it is doing an
inefficient sort.  This goes on for 19 hours, just to create empty directories.
Then it can work on restoring real files.

       Mike (not ADSM support) Knight