07-06-07, 12:06 #1Registered User
- Join Date
- Mar 2005
Unanswered: performance/monitoring question...
Running our software on many client sites.
1) using a 4gl language
2) uses ntwdblib.dll
Client MSSQL Servers:
1) varies from site to site on average dual core, 4GB, 20-100GB HD
2) MSSQL 2000 sp3-4, 2005 sp1-2
A handfull of our clients have complained about how slow things seem to be working. So after getting online with the customer and doing some investigating/testing, I have found that 'slow' is not the word to use. We are talking about snails pace. Now this same software at our other mssql sites doesnt seem to have any problems. very good performance numbers. Some of our customers with single cpu, 2GB, 10GB HD have better performance than a couple of our 4 cpu, 16GB, 2 - 20GB HD systems.
we have written a simple test program that we run on these slow sites that individually (useing local drives/files) are comparable to our faster sites. We then run the test program against the MSSQL database/tables. On the servers the run times are about the same. However, when we run it on a workstation the results go off the scale. At our good sites, the test time is around 10 minutes, on our bad sites the test time is around 1 HR 30 minutes.
Now i know this sounds like a network issue, but one of these slow sites had someone come out and check the network traffic/packets/routers/switches. then we paid to send that person to one of our faster sites to test their network. The results were that our slower client had a faster network. Our slower client uses CISCO smart switches/hubs/router.
Now for my question, does anyone know of or could recommend something that i could use to determine what is happening between the server and workstation on these slow sites.
Is there a problem with CISCO equipment and MSSQL 2005, NTWDBLIB.DLL. I am looking for anything here.
07-06-07, 12:14 #2Annie's Dog Walker
Provided Answers: 6
- Join Date
- Nov 2004
- on the wrong server
i would start with sql server's profiler and the windows performance monitor to rule in\out sql server as the issue. there is plenty of documentation on these tools in Books Online and elsewhere.If one brings so much courage to this world the world has to kill them or break them, so of course it kills them. The world breaks every one and afterward many are strong at the broken places. But those that will not break it kills. It kills the very good and the very gentle and the very brave impartially. If you are none of these you can be sure it will kill you too but there will be no special hurry. Earnest Hemingway, A Farewell To Arms.
07-06-07, 12:31 #3Registered User
- Join Date
- Mar 2005
... we have run the test program
on the MSSQL servers against the database/tables configuration of our 4GL. ON our single cpu sites and the program runs in about 4-5 minutes on average, when I ran the test program on our 4 cpu sites the program finishes in about 3-4 minutes average time. We have ran the monitors on both sites to see if anything jumped out at us. We basically confirmed that 4GL is not very good at speed/resource issues which we already new(page splits being our hardest hit indicator). But the results showed that the 4 cpu systems outperformed the single cpu systems on a head to head run.
FYI: With our 4GL we can configure the test program to use non-sql mode or sql mode. non-sql mode uses flat files on local or network drives. sql-mode of course uses a table in a database. on the 4 cpu servers and single cpu servers, when i run the test, i have configured it both ways just to see if anything stood out. the 4 cpu systems pretty much walked all over the single cpu system.