NLM and reporting performance (OpenInsight Specific)
At 15 SEP 1998 01:35:11PM Paxton Scott (ARCS, Inc.) wrote:
Netware 3.11, server all updates, pentium 120 32mb, ipx/spx 802.3
NLM 1.5
Both AREV 3.12 and OIWG 3.61 are showing FFFFFF in file handles
REVPARAM files in all rev subdirectories.
Seems to work OK.
Concerened about report performance.
The report, basically RLIST (ORMAIN) has 5 selections on 2 fields both indexed. Table is 33,000 records
Running from local files on the work station, takes less than 3 min.
Running on the network as described above, about 34.5 min
On network W/O NLM between 30 and 40 mins. (Wasn't watching that closly by then!) Not much difference.
Are these times reasonable? What might I do to tune this test bed here before I recommend to the client about more server hardware, etc…
……………..
Thanks
At 24 SEP 1998 10:39PM Cameron Revelation wrote:
Paxton,
]] Running from local files on the work station, takes less than 3 min.
That is because the files are cached as much as possible. The disk is probably be used minimally. You can see this with "Cache Hits" stats and stuff like that in Windows system tools monitor.
]] Running on the network as described above, about 34.5 min
That is an amazing difference. Is the server "underpowered" compared to the workstation? What does the server utilization look like?
]] On network W/O NLM between 30 and 40 mins. (Wasn't watching that closly by then!) Not much difference.
Using the NPP with only one workstation? If so, that caches the files locally as well.
Cameron Purdy
Revelation Software
At 02 OCT 1998 07:41AM Paxton Scott wrote:
Thanks, Cameron, for your reply. I do think my customer has an under powered server, but I was trying to 'prove' that before I told him to upgrade it. I'd like to be sure he'll see some improvements. And, based on my tests on my system, I'm not comfortable that that would happen. NOw, response to your questions about my 'test bed'
Well, basically the only job on the server., will try to look at usage, but it has 32 meg and only 4 or five other idle clients.
The client running the reports is faster, 266Mh, 128MB but it still I'm not useing the NPP, only the NLM. Still, It seems like it should get better performance. I must be missing something. Any tests you can think of that I might try?