Slow File Transfer

[Guest]

I am currently connected to a small business network that has one server
(Server 2003) and four other computers (2000 and XP Pro).

When I am logged onto the server, I can transfer files to and from every
computer just fine except for one (which I'll call computer1). When I try to
copy files from computer1 to the server, while still logged onto the server,
the files copy very slowly, but I can copy from the server to computer1 just
fine. Funny thing is that when I'm logged onto computer1, or any other
computer on the network, I can copy files to/from the server and every other
computer without a problem. The only slow file transfer occurs when I am
logged onto the server and copying from computer1 to the server.

Any takers?

 

[Chuck]

I am currently connected to a small business network that has one server
(Server 2003) and four other computers (2000 and XP Pro).

When I am logged onto the server, I can transfer files to and from every
computer just fine except for one (which I'll call computer1). When I try to
copy files from computer1 to the server, while still logged onto the server,
the files copy very slowly, but I can copy from the server to computer1 just
fine. Funny thing is that when I'm logged onto computer1, or any other
computer on the network, I can copy files to/from the server and every other
computer without a problem. The only slow file transfer occurs when I am
logged onto the server and copying from computer1 to the server.

Any takers?

Slow file transfers are a lot of fun to diagnose. Your description of the
problem is a bit hard to follow though.

There are two directions the data can move.
1) From Computer1 to the Server.
2) From the Server to Computer1.

There are two ways to run the transfer.
1) While sitting in front of Computer1.
2) While sitting in front of the Server.

This gives you 4 tests to run.
1) Pushing to Computer1, sitting at the Server.
2) Pushing to the Server, sitting at Computer1.
3) Pulling from Computer1, sitting at the Server.
4) Pulling from the Server, sitting at Computer1.

There are various ways to move the data.
1) Using FTP.
2) Using drag and drop (Windows Explorer).
3) Using the Copy or XCopy commands from a command window.

I'm betting that you're using Windows Explorer. I think that you've discussed
moving the data in both directions. I'm not what "logged onto" means though.
Do you mean which computer you're sitting in front of, running the desktop of?
Or connected to,thru the network?

Now, if the slow transfer ONLY affects Computer1, then it's a good bet that the
problem originates with something about Computer1. But there are a boatload of
possible factors, and any of them can have an asymmetrical effect.
* Network connection. Network cabling, router / switch port, network card.
* Hardware. CPU, disk.
* Software. What's running on it, that you know of?
* Disk fragmentation.
* Security, and network use. What could be running on it, that you don't know
of?

How similar is Computer1 to all other computers that you've tested this problem
on, and to your network environment as a whole, given the above factors?

 

[Guest]

Slow file transfers are a lot of fun to diagnose. Your description of the
problem is a bit hard to follow though.

There are two directions the data can move.
1) From Computer1 to the Server.
2) From the Server to Computer1.

There are two ways to run the transfer.
1) While sitting in front of Computer1.
2) While sitting in front of the Server.

This gives you 4 tests to run.
1) Pushing to Computer1, sitting at the Server.
2) Pushing to the Server, sitting at Computer1.
3) Pulling from Computer1, sitting at the Server.
4) Pulling from the Server, sitting at Computer1.

There are various ways to move the data.
1) Using FTP.
2) Using drag and drop (Windows Explorer).
3) Using the Copy or XCopy commands from a command window.

I'm betting that you're using Windows Explorer. I think that you've discussed
moving the data in both directions. I'm not what "logged onto" means though.
Do you mean which computer you're sitting in front of, running the desktop of?
Or connected to,thru the network?

Now, if the slow transfer ONLY affects Computer1, then it's a good bet that the
problem originates with something about Computer1. But there are a boatload of
possible factors, and any of them can have an asymmetrical effect.
* Network connection. Network cabling, router / switch port, network card.
* Hardware. CPU, disk.
* Software. What's running on it, that you know of?
* Disk fragmentation.
* Security, and network use. What could be running on it, that you don't know
of?

How similar is Computer1 to all other computers that you've tested this problem
on, and to your network environment as a whole, given the above factors?

I'll do my best to map out the situation a bit more for you. First of all,
I'm not physically sitting at the server, I'm logged onto it through Remote
Desktop Connection, since there's no monitor connected to the server. But
(please correct me if I'm wrong) that should be exactly the same as sitting
at the server. None of the computers are the same, although some of them run
the same OS (XP Pro).

Here are the transfer patterns(all using explorer):

1) Sitting in front of Computer1
Pushing to server - Good
Pulling from server - Good
Pushing to any other network computer - Good
Pulling from any other network computer - Good

2) Sitting in front of Computer1, but logged onto server through Remote
Desktop
Pushing to computer1 - Good
Pulling from computer1 - BAD!!!
Pushing to any other network computer - Good
Pulling from any other network computer - Good

3) Sitting in front of any other computer
Pushing to server - Good
Pulling from server - Good
Pushing to any other network computer - Good
Pulling from any other network computer - Good

4) Sitting in front of any other computer, but logged onto server through
Remote
Desktop
Pushing to computer1 - Good
Pulling from computer1 - BAD!!!
Pushing to any other network computer - Good
Pulling from any other network computer - Good

Thank you for your quick response!!!

 

[Chuck]

I'll do my best to map out the situation a bit more for you. First of all,
I'm not physically sitting at the server, I'm logged onto it through Remote
Desktop Connection, since there's no monitor connected to the server. But
(please correct me if I'm wrong) that should be exactly the same as sitting
at the server. None of the computers are the same, although some of them run
the same OS (XP Pro).

Here are the transfer patterns(all using explorer):

1) Sitting in front of Computer1
Pushing to server - Good
Pulling from server - Good
Pushing to any other network computer - Good
Pulling from any other network computer - Good

2) Sitting in front of Computer1, but logged onto server through Remote
Desktop
Pushing to computer1 - Good
Pulling from computer1 - BAD!!!
Pushing to any other network computer - Good
Pulling from any other network computer - Good

3) Sitting in front of any other computer
Pushing to server - Good
Pulling from server - Good
Pushing to any other network computer - Good
Pulling from any other network computer - Good

4) Sitting in front of any other computer, but logged onto server through
Remote
Desktop
Pushing to computer1 - Good
Pulling from computer1 - BAD!!!
Pushing to any other network computer - Good
Pulling from any other network computer - Good

Thank you for your quick response!!!

Kewl. RDC is another factor that I considered briefly, that will make this even
more fun. Any remote control or remote desktop imposes load upon the computers
involved, and different products, different factors.

Very nice canonical problem analysis, above. We see two problem test cases,
with all other 14 test cases being no problem.
* Computer1 as RDC client, to server as RDC server, pulling from Computer1
* Computer2 as RDC client, to server as RDC server, pulling from Computer1

My gut feeling is that this clearly indicts "pulling from Computer1" to be the
common factor. I think it exonerates RDC (both client and server processes) as
being involved. And it exonerates the Server as being involved.

Now is when you will need Task Manager, or maybe (preferably) Process Explorer,
running on Comupter1, to see which resource is being overloaded. Both CPU,
disk, network card can be easily monitored under either Task Manager or Process
Explorer.

And what is the homogeneity factor - Computer1 to Computer2? Hardware,
software, network use?

Any time I had any problem involving any disk related activity, I would check
(correct) disk fragmentation first though. Have you done this?

 

[Guest]

Kewl. RDC is another factor that I considered briefly, that will make this even
more fun. Any remote control or remote desktop imposes load upon the computers
involved, and different products, different factors.

Very nice canonical problem analysis, above. We see two problem test cases,
with all other 14 test cases being no problem.
* Computer1 as RDC client, to server as RDC server, pulling from Computer1
* Computer2 as RDC client, to server as RDC server, pulling from Computer1

My gut feeling is that this clearly indicts "pulling from Computer1" to be the
common factor. I think it exonerates RDC (both client and server processes) as
being involved. And it exonerates the Server as being involved.

Now is when you will need Task Manager, or maybe (preferably) Process Explorer,
running on Comupter1, to see which resource is being overloaded. Both CPU,
disk, network card can be easily monitored under either Task Manager or Process
Explorer.

And what is the homogeneity factor - Computer1 to Computer2? Hardware,
software, network use?

Any time I had any problem involving any disk related activity, I would check
(correct) disk fragmentation first though. Have you done this?

Well, I had already checked the resources a couple days ago and nothing is
being overloaded. Network, CPU, and RAM are all good to go. As for software,
I'm certain it's a software issue. I haven't checked the fragmentation, and I
guess I'm not sure how that would affect pulling from the computer with the
server, since pulling with other computers yields a good transfer rate, but I
will definately check that when I get back to the office on Monday.
Personally, I'm convinced that it's a software issue, but I have no idea
where to start diagnosing networking software.

 

[Chuck]

Well, I had already checked the resources a couple days ago and nothing is
being overloaded. Network, CPU, and RAM are all good to go. As for software,
I'm certain it's a software issue. I haven't checked the fragmentation, and I
guess I'm not sure how that would affect pulling from the computer with the
server, since pulling with other computers yields a good transfer rate, but I
will definately check that when I get back to the office on Monday.
Personally, I'm convinced that it's a software issue, but I have no idea
where to start diagnosing networking software.

You start by finding the bottleneck. Any time you transfer a file, and the file
transfer takes more than 0 seconds, then there is some resource somewhere that
is being exhausted, and slowing the file transfer down. So, you find the
resource that's being exhausted. That resource will show at 100% utilisation
while the file transfer is going.

And that's the value of Process Explorer.
<http://nitecruzr.blogspot.com/2006/04/watching-what-your-computer-is-doing.html>
http://nitecruzr.blogspot.com/2006/04/watching-what-your-computer-is-doing.html

But, you never told us - is Computer1 identical to the others? Hardware,
software, usage pattern? This is highly relevant. If there's anything about
Computer1, that makes it less powerful or more burdened than the others, now
would be a good time to discuss it. How homogenous is your LAN?

 

[Guest]

You start by finding the bottleneck. Any time you transfer a file, and the file
transfer takes more than 0 seconds, then there is some resource somewhere that
is being exhausted, and slowing the file transfer down. So, you find the
resource that's being exhausted. That resource will show at 100% utilisation
while the file transfer is going.

And that's the value of Process Explorer.
<http://nitecruzr.blogspot.com/2006/04/watching-what-your-computer-is-doing.html>
http://nitecruzr.blogspot.com/2006/04/watching-what-your-computer-is-doing.html

But, you never told us - is Computer1 identical to the others? Hardware,
software, usage pattern? This is highly relevant. If there's anything about
Computer1, that makes it less powerful or more burdened than the others, now
would be a good time to discuss it. How homogenous is your LAN?

I did actually mention a couple replies ago that none of the computers on
the network are the same. There are two PC's, one running XP Pro, the other
Win 2000, and there are two laptops. As for resources, I don't know the
hardware setup for the separate comptuers, since I never use them. But
Computer1 (which is the computer I use) does have plenty of resources. My
boss mentioned that he's ordered some spare network cards, so we will try
swapping out the NIC as soon as those come in, but before then I will
definately look into the resources again. I'll let you know what I find.

 

[Chuck]

I did actually mention a couple replies ago that none of the computers on
the network are the same. There are two PC's, one running XP Pro, the other
Win 2000, and there are two laptops. As for resources, I don't know the
hardware setup for the separate comptuers, since I never use them. But
Computer1 (which is the computer I use) does have plenty of resources. My
boss mentioned that he's ordered some spare network cards, so we will try
swapping out the NIC as soon as those come in, but before then I will
definately look into the resources again. I'll let you know what I find.

OK, well if I missed a detail which you provided, I apologise. But what you
have to do is find the bottleneck. There is one, somewhere.

I presume that, when you test Computer1 vs Computer2, you are moving the same
file(s) in each test? And you're using the same procedure / utility to move the
file(s) in each test?

If Process Explorer, which shows just 3 metrics - CPU utilisation, memory
utilisation, and I/O, isn't detailed enough, you have (Control Panel -
Administrative Tools - ) Performance. Performance Monitor, aka System Monitor,
offers dozens of metrics that you can watch, on any NT based computer. If any
resource doesn't reach 100% utilisation, then it's not the (only) problem. Look
at all metrics, until you find the problem.

 

[Guest]

OK, well if I missed a detail which you provided, I apologise. But what you
have to do is find the bottleneck. There is one, somewhere.

I presume that, when you test Computer1 vs Computer2, you are moving the same
file(s) in each test? And you're using the same procedure / utility to move the
file(s) in each test?

If Process Explorer, which shows just 3 metrics - CPU utilisation, memory
utilisation, and I/O, isn't detailed enough, you have (Control Panel -
Administrative Tools - ) Performance. Performance Monitor, aka System Monitor,
offers dozens of metrics that you can watch, on any NT based computer. If any
resource doesn't reach 100% utilisation, then it's not the (only) problem. Look
at all metrics, until you find the problem.

Yes, all the same files are being moved with the same utilities.

Interestingly enough...the RAM appeared to be the bottleneck, which
shouldn't slow anything down since the server has 1 GB and Computer1 has 1.25
GB. I used System Monitor to look at the resources. I noticed that, when on
the server pulling from Computer1, the RAM usage on the server will peak at
100% and then fall down to 0%, and it will continue doing that over and over
until the copy is done. When using the server to pull from any other
computer, the RAM on the server peaks at 100% and then remains peaked until
the copy is done.

 

[Chuck]

Yes, all the same files are being moved with the same utilities.

Interestingly enough...the RAM appeared to be the bottleneck, which
shouldn't slow anything down since the server has 1 GB and Computer1 has 1.25
GB. I used System Monitor to look at the resources. I noticed that, when on
the server pulling from Computer1, the RAM usage on the server will peak at
100% and then fall down to 0%, and it will continue doing that over and over
until the copy is done. When using the server to pull from any other
computer, the RAM on the server peaks at 100% and then remains peaked until
the copy is done.

Kewl. Now we're getting to the fun part. You're describing RAM on the server,
in both cases?

So it shouldn't be too hard to find the key factor on Computer1. How many other
computers? What one factor on Computer1 is different from all of the other
computers put together? With lack of homogeneity (as you describe), and all of
the other test cases that don't have a problem, it shouldn't be too hard to find
the one factor that's unique to Computer1.

 

[Guest]

Well, I would like to thank you for your help, but unfortunately I've been
forced to put this problem aside for now, and I have no idea when I'll be
able to get back to it. I may have a chance in the next week or two to swap
out the network card, however, and I will do my best to remember to post
another reply if I'm able to solve the problem. Again, thanks for all the
help Chuck!!!

 

[Guest]

I am currently connected to a small business network that has one server
(Server 2003) and four other computers (2000 and XP Pro).

When I am logged onto the server, I can transfer files to and from every
computer just fine except for one (which I'll call computer1). When I try to
copy files from computer1 to the server, while still logged onto the server,
the files copy very slowly, but I can copy from the server to computer1 just
fine. Funny thing is that when I'm logged onto computer1, or any other
computer on the network, I can copy files to/from the server and every other
computer without a problem. The only slow file transfer occurs when I am
logged onto the server and copying from computer1 to the server.

Any takers?

BAM! I have no idea why, but swapping the NIC fixed it. I'm kind of confused
as to how a network card can interface differently like that, but oh well.
The problem is fixed! If anyone's got any theories or anything, I'd be
interested to hear them. I figure it's a good learning experience. Thanks for
the help Chuck!!!