Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
rlaubert
Yak Posting Veteran
96 Posts |
 Posted - 2007-03-14 : 13:35:07
|
| Have a new server and application in testing. This app is a documet imaging system using SQL 2000 and currently in a 10% pilot (10 percent of our daily documents are scanned into it for testing).I am getting some reading with performance monitor that I am not sure of.Page Faults/Sec 155 -10120 (min-max) avg 414Page /sec 0 - 1530 (min-max) avg 15% disk Read 0%% disk Write 0-139 avg 4% disk Time 0-139 avg 4Disk % are for the c drive only the other drives are 0%Disk Queue 0-2 avg 0% CPU <2% Max 13%These reading are over about 4 min time frame but are pretty consistent.Major activity atm is scanning and importing documents (tif files)My concern is the amount of page faults, isn't this excessive?Several of my other production servers also have high page faults so I am interested in seeing what the experienced admins think.ThanksRaymond LaubertMCDBA, MCITP:Administration, MCT |
|
|
snSQL
Master Smack Fu Yak Hacker
1837 Posts |
Posted - 2007-03-14 : 13:51:40
|
| See http://www.microsoft.com/technet/prodtechnol/sql/2000/maintain/rdbmspft.mspx |
 |
|
|
rlaubert
Yak Posting Veteran
96 Posts |
Posted - 2007-03-15 : 10:25:14
|
| snSQL,I am well aware of that page and how to monitor SQL. What I was interested in was some insight from people that have experience in production servers and the readings I am getting from mine.With high page faults and sort of high pages/sec (>10) it seems to indicate an issue. Hence the reason for the post is to find areas to look at.Raymond LaubertMCDBA, MCITP:Administration, MCT |
|
|
|
Michael Valentine Jones
Yak DBA Kernel (pronounced Colonel)
7020 Posts |
Posted - 2007-03-15 : 10:58:47
|
| You have to look at the type of page faults you are getting: cache faults, demand zero faults, page read/sec, page writes/sec, pages input/sec, pages output/sec.The faults could file system cache faults, which is an indication of file system activity. Those could just be database backups or transaction log backups.CODO ERGO SUM |
|
|
|
snSQL
Master Smack Fu Yak Hacker
1837 Posts |
Posted - 2007-03-15 : 12:06:58
|
| I assumed them to be hard and soft memory page faults (Raymond, you didn't specify which page faults counter you are looking at), which could mean you need more memory, or at least you need to have less applications contending for the meory in the server. |
|
|
|
rlaubert
Yak Posting Veteran
96 Posts |
Posted - 2007-03-15 : 14:53:21
|
| Been looking into this more today, seems like I have a contention for memory between the OS, other applications and SQL. SQL was installed using default settings by the staff in FL. The box has 4GB if RAM. I did limit SQL to 2048 since that is all SQL was using anyway. No /3GB in configuration settings. But don't think this will help. I did find that McAfee was running in real time (have ticket in to disable realtime mode). Server has 3 other processes related to this application: fax, remote server (admin), and filenet (old application being converted).However, during the monitoring period none of these showed any activity other than virus scan.Will just have to keep looking. Info has to be here somewhere.These were memory/page faults/sec and memory/pages/sec entries.Raymond LaubertMCDBA, MCITP:Administration, MCT |
|
|
|
|
|
|
|
|
Disk issue?