It is slow over the network for one. The server itself seems to be slow to respond but only when there are lots of people at work (which I expected would cause some performance decreases).
It functions as an App server, DNS, DHCP, WINS, file, printer and a DC. Yes it does a lot, but it has been doing this for a while. I still am led to believe that a better design for our network would do some good stuff for performance on this server as well as the other main File and Email server on our network.
I'd say the main two issues I am trying to resolve are the slow response times from the Application and the lag in response I get from the File server (mainly it just hangs when you try to access it and then after a minute will be very quick as if it was waiting in line on the network).
Hope this helps and I am going to see if I can use the performance monitor to identify the problem, but I am not the best understanding what is overloading and what is within normal functionality, guess I'll just have to try it and learn it.
It's not a war on drugs it's a war against personal freedoms!
Here is a basic segment that I was monitoring. Lots of spikes from the HD, but should I be looking at anything specifically that would shed more light on where the problem lies...or perhaps I am staring straight at it and not knowing it.
On the graph I uploaded it shows the Avg. Disk Queue Length in green, Pages/sec of memory in dark blue, packets/sec in light blue/teal, processor in red and bytes total/sec in pink.
I don't know if you can make it all out, but perhaps this will give some insight.
It's not a war on drugs it's a war against personal freedoms!
Ok so just over the last few minutes every part of the performance monitor was peaking, except the processor. It seems the server is just running at full load constantly.
Just a moment after I typed the top portion the monitor started to calm down, but for a good few minutes everything was maxing out. Packets, bytes and other related network monitoring are still maxing out. The HD, Processor, memory are all more or less back to normal again.
It's not a war on drugs it's a war against personal freedoms!
yeah, things we have considered to reduce access loads:
1) moving shared printers
2) moving desktop icons etc to local pc
3) moving start menu to local pc
4) reducing logs
5) removing AV
Unfortunately for us none of those is an option right now, but maybe for you.
If none of those helps/is possible its time to get a second server to share the load.
If the world doesn't stop annoying me I will name my kids ";DROP DATABASE;" and get revenge.
So one area I have already begun replacing is all the Hubs in our office with Switches. This should help out since I was detecting lots of RX and TX errors from people who have a lot of nodes at their desks and were forced to use Hubs from a while back.
Another thing I am going to do is use a Layer 3 switch to give priority to the main server.
I don't believe the server has AV on it, but I'll double check. I should find a way to move the print server off their as well, but it isn't used very much anyways. What logs do you speak of? I think that would also help, but not sure what logs are taking place on this server. I don't believe there are any roaming profiles, but I'll do a quick double check to be sure.
It is too bad no one here is a Perforce expert, but I am learning it as I go. Do you think if I could somehow offload some of the load onto a secondary network card in the server that it might help? Unfortunately these servers were all set up as is long before I got here.
It's not a war on drugs it's a war against personal freedoms!
Replacing those hubs should be priority one. You mentioned your network traffic has increased recently and those hubs are allowing a lot of collisions.
Once you have your switches in place configure your network cards to run in full duplex. This should make a huge difference.
Thanks for confirming my thoughts. These hubs were causing a lot of collisions. Now that we are putting the switches in place things should get a bit better.
So are you saying setting the NICs to 100Mb/Full Duplex mode instead of auto is a better method? What about the other option of setting it to auto negotiate/1000Mb.
It's not a war on drugs it's a war against personal freedoms!
If your hardware is capable of 1Gb/s speeds configure it as such. Hardcoding the nic and switch to full duplex is recommended as this essentially doubles your bandwidth and if you can do so you should with the speed at 1Gb/s. You could leave it to auto-negotiate although this is typically frowned upon for reasons I am unsure but apparently having nics operating at varying duplex is a bad thing.
Ignoring the details replacing those hubs irregardless of your nics configuration will prove a drastic improvement.