| Title: | *OLD* ALL-IN-1 (tm) Support Conference |
| Notice: | Closed - See Note 4331.l to move to IOSG::ALL-IN-1 |
| Moderator: | IOSG::PYE |
| Created: | Thu Jan 30 1992 |
| Last Modified: | Tue Jan 23 1996 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 4343 |
| Total number of notes: | 18308 |
A customer has indicated that ever since their V2.3-V3.0 upgrade, ALL-IN-1 response time has greatly degraded. After doing some performance monitoring with DECps, they have found that the System Fault Rate appears to be excessively high, which it wasn't before while at V2.3. Since V3.0 during monitoring, It peaked at 24. The recommended value for this is anything below 1. They have also noticed that the requirement for the RESHASHTBL has doubled. They had to increase it from 4096 to 10000. They are wondering if the above behaviour is in line with expections when upgrading to V3.0. We are not expected to do any tuning on their system but this customer will absolutely not let go until I can that say someone has either confirmed or denied this. They are at the point now that they feel that something went wrong in the install (something is not linked in right) even though nothing did. VAX 4000/500 100+ active users in ALL-IN-1, and sub-processed out to other applications. 192 Meg of memory System scattered accross 5 disks. /Mario
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 3439.1 | This has not been the experience of my customers | GIDDAY::SETHI | Holland 2-England 0,Andrew wasn't there | Tue Oct 26 1993 03:38 | 18 |
Hi Mario,
It appears that they have to go through a tuning exercise as this has
not been our customers experience here in Australia. In general people
have been quite pleased even though ALL-IN-1 requires additional
resources.
It's all very well using DECps but people should not just depend on it.
The customer has other applications running therefore a number of
factors may come into play.
You are not going to get absolute answers so please don't take my reply
to be The absolute answer, even though I am attempting to be an
explainator :-).
Regards,
Sunil
| |||||
| 3439.2 | possible explanations | IOSG::DAVIS | Mark Davis | Tue Oct 26 1993 08:12 | 31 |
Some performance degradation is to be expected when moving from V2.4 to
V3.0.
You would expect to see on average 2% more CPU to be used and 8% more
memory. If a lot of printing takes place using the WPS-PLUS formatter
then the degradation will be worse.
You would not expect response time to be greatly degraded. Possible
explanations are :-
You are already at the limit of CPU or memory so that the extra usage
pushes you over the edge.
User behaviour has changed because of the extra capabilities of V3.0 so
that users are doing more work (?).
Something changed during the upgrade - for example global buffers on
PENDING.DAT and the SDAFs.
As for the increase in RESHASHTBL, although I haven't observed this,
the explanation could be that V3.0 opens more files and therefore
requires more locks. - it is usually recommended that RESHASHTBL is a
power of 2.
Maybe the increase in the system fault rate reflects a general increase
in page faulting.
Mark
| |||||