Performance issue with NT Server

[tgilmore]

Hello. I have a NT Server in a single domain with very few users. Performance has never been an issue until just recently. We have an application running on the server that the end users use throughout the day and the performance of this application has degraded significantly. Also, the users documents are stored on the server. When a user accesses their Word documents across the network it takes quite awhile for the document to come up. Nothing has changed with this server and there is plenty of HD space. I have defragged the drive and it didn't seem to improve the performance any. There are not any events in the logs either. Any idea's?

Thank you in advance.

TGilmore

 

[SgtB]

Possible memory leak in the application you're using....or some other app. Just a theory. Open up the task manager and sort by memory use. See whats taking the most memory. See how much memory you have available. Have you rebooted lately?

What was the last thing done to the server when the problems started? Was the performance problem gradual or instant?

It really sounds like a memory/cpu problem. Check the task manager, and find out whats bogged down.

Hope this helps

 

[hjm3857]

I agree about the memory leak being a good possibility, but since you described the primary perfomance problem being with accessing a file on the server and not with the performance of the application on the box, I also suspect Disk and network performance. You have gone down the correct path with a defrag, but it is good to check the disk IO performance with perfmon, (you will need to enable diskperf if it is not already done. Find instructions by typing "diskperf" at the command line.) you will need to look at a counter under physical (or logical) disk called current disk queue length and average disk queue length. If this counter is higher that 2 or 3 you have a bottleneck with you disk IO subsystem. If the server has an array controller you may be able to tune the caching to add more read cache to help, or you may need to change hardware. As a note if your server is also running SQL be careful with array cache settings as write cache being enabled (the default) can corrupt your database.

If your disk IO looks OK, check the network port you are connected to. If it is a switch port (hopefully) you are best swerved if you can hard set to the highest possible common setting (hopefully 100 meg full duplex) (careful when making a change here as it may temporarily disconnect your server. If your nic and port are at autosense (the default) you can check the switch port for error rates and identify the problem before you start making changes.


As a final note, I assumed that before you started to dig into the server as a problem you tried to access the file from multiple other locations in the network to make sure it is not an issue with the workstation.


I hope this helps,
Jay Mosser
jaym@optymgroup.com

 

[JimInKS]

sgTB and htm3857 both have good suggestions.

If you have a managed switch look at the port stats. Also look at the "Network" tab in "Windows NT Diagnostics" and see if you see a large number of errors.

Also, don't know what kind of app you are running, but if it is some kind of a database there may be some administrative functions you can run to re-index or otherwise clean the db up. It could also be the case that you just have a lot more data than you used to and it taking longer to access.

 

[hjm3857]

FYI for JimInKS

Be careful when looking as the stats for errors on your interface. Most NIC drivers (3COM. Intel, etc.) for the MS world don't collect error or collision stats and that will make the stats look clean on the server side when they may actually be bad. I have always found it best to stick with the switch port stats.

 

[tgilmore]

Thank you all for your reply's. I know that it is not a memory leak in the application because the server/network performance degrades everytime a file is accessed from the server. For example, when they access a Word document it takes forever to load, save and close the document and at times it generates errors. The configuration for the server has not changed at all since this has been happening it just all of the sudden happened. I rebooted the machine last night. I believe my problem lies in the network itself. I have checked perfmon and the counter was at 0 for physical and logical disks. I inspected our hub and there were not very many collisions and all of the indicator lights appeared to be working properly. In NT Diagnostics there were a few errors under the network tab. In Task Manager the memory hog is rtvscan.exe then vptray.exe. I'm pretty sure that these processes are related to Norton Antivirus but another strange thing is that the real-time virus protection is disabled. Any help would be greatly appriciated.
TGilmore

 

[JimInKS]

hjm3857 Thanks, I wasn't aware of that.

tgilmore -
If you have 3com NICs (others may do this also) you can usually run a network test between two PC's, your server and one of the workstations. This will test your network hardware and cabling.

Go to Network properties and look for "diagnostics". I don't remember all the details but you can set one pc up as a "receiver" or "echo" device. Then the other PC can initate a diagnostic conversation with the "echo" PC that will test the integrity of the connection between the two.

There may also be a DOS version of this with the 3com install disks.

Note: Running this test will drop your windows network connection.