Got a new gigabit switch

[DisgruntledVirus]

Got a gigabit switch off the fs/ft forum and plugged it in to test all ports. It worked, but didn't test network speeds or anything. So went ahead and redid some cabling of the network so everything was on the gigabit switch. Booted up Warcraft 3 to play online with some friends after I was done, and didn't notice any issues. Started playing and got choppiness like mad. It was completely unplayable. So, I tried another game thinking it was that host. Nope, same thing. Tried L4D, and same thing. Then tried the internet and took 10+ seconds to load google.com.

Well, did a basic trace route, and it looks like it's dropping packets when there are other devices plugged in. First trace route is the switch, the 2nd is the switch again but this time with only my desktop plugged in (no server or anything to eliminate any possible outside issues), and the third is plugged directly into the router. All were done with uplink on port 1.

C:\Documents and Settings>tracert google.com

Tracing route to google.com [74.125.127.100]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.1.1
2 9 ms 10 ms 6 ms 10.71.224.1
3 7 ms 15 ms 8 ms gig15-1.ncntoh1-rtr2.neo.rr.com [24.164.102.173]

4 * 14 ms 14 ms tge2-0-1.clevoh1-rtr0.mwrtn.rr.com [65.25.137.23
7]
5 17 ms 18 ms 19 ms ae-3-0.cr0.dca20.tbone.rr.com [66.109.6.70]
6 52 ms 18 ms 18 ms ae-1-0.pr0.dca10.tbone.rr.com [66.109.6.165]
7 * 20 ms 19 ms 66.109.9.66
8 30 ms 18 ms 21 ms 216.239.48.112
9 54 ms 43 ms 43 ms 209.85.242.208
10 67 ms 60 ms 57 ms 209.85.249.19
11 98 ms 113 ms 101 ms 209.85.241.211
12 100 ms 97 ms 97 ms 216.239.46.212
13 99 ms * 101 ms 64.233.174.99
14 104 ms 106 ms 107 ms 216.239.46.2
15 95 ms 100 ms 94 ms pz-in-f100.google.com [74.125.127.100]

Trace complete.

C:\Documents and Settings>tracert google.com

Tracing route to google.com [74.125.127.100]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.1.1
2 22 ms 10 ms 6 ms 10.71.224.1
3 16 ms 8 ms 17 ms gig15-1.ncntoh1-rtr2.neo.rr.com [24.164.102.173]

4 9 ms 19 ms 9 ms tge2-0-1.clevoh1-rtr0.mwrtn.rr.com [65.25.137.23
7]
5 18 ms 19 ms 17 ms ae-3-0.cr0.dca20.tbone.rr.com [66.109.6.70]
6 45 ms 59 ms 22 ms ae-1-0.pr0.dca10.tbone.rr.com [66.109.6.165]
7 17 ms 18 ms 18 ms 66.109.9.66
8 22 ms 20 ms 17 ms 216.239.48.112
9 85 ms 48 ms 49 ms 209.85.242.208
10 64 ms 65 ms 65 ms 209.85.249.19
11 103 ms 103 ms 104 ms 209.85.248.129
12 98 ms 100 ms 112 ms 216.239.46.204
13 100 ms 98 ms 105 ms 64.233.174.97
14 97 ms 108 ms 108 ms 216.239.46.2
15 102 ms 106 ms 106 ms pz-in-f100.google.com [74.125.127.100]

Trace complete.

C:\Documents and Settings>tracert google.com

Tracing route to google.com [74.125.127.100]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.1.1
2 8 ms 18 ms 7 ms 10.71.224.1
3 8 ms 10 ms 8 ms gig15-1.ncntoh1-rtr2.neo.rr.com [24.164.102.173]

4 10 ms 10 ms 10 ms tge2-0-1.clevoh1-rtr0.mwrtn.rr.com [65.25.137.23
7]
5 19 ms 19 ms 19 ms ae-3-0.cr0.dca20.tbone.rr.com [66.109.6.70]
6 30 ms 17 ms 17 ms ae-1-0.pr0.dca10.tbone.rr.com [66.109.6.165]
7 17 ms 18 ms 20 ms 66.109.9.66
8 18 ms 35 ms 18 ms 216.239.48.112
9 43 ms 43 ms 41 ms 209.85.242.208
10 65 ms 68 ms 67 ms 209.85.249.19
11 104 ms 127 ms 103 ms 209.85.248.129
12 107 ms 115 ms 105 ms 216.239.46.204
13 106 ms 105 ms 114 ms 216.239.48.143
14 119 ms 114 ms 105 ms 216.239.46.22
15 96 ms 106 ms 106 ms pz-in-f100.google.com [74.125.127.100]

Trace complete.

Update: Called DLink to see about an RMA. Ultimately ended up talking to a level 3 tech support guy that was very helpful. After talking about it with him, he said it sounds environmental(i.e. my network specifically and how it's setup) which the symptoms point to (if we assume the previous owner didn't have any issues with it). So, he asked if I could test it using a different environment (which is fair enough), so I will be using it on a different network and run the same tests on the other network.

While it's not a new switch, I think it's fair for me to test it in another network based on the results I'm seeing before they RMA it.

 

[Nuwave]

Is this a managed switch? What model is it?

 

[JackMDS]

What ever switch is used 100MB/sec. or Giga switch should not make any difference waht so ever for playing on line.

You first line of the trace is from the computer to the Router and it is always 1MS as it should.

In all instances at the second trace you are already out on the Internet.

Unless you played with the Computer's TCP/IP setting and sc*w some thing up, or you have a manged switch that set incorrectly, this should not happen.

Measuring the performance of Network Devices on a LAN via Gaming Server on the Internet is "Silly" at Best.

You want to evaluate your switch and local computers, http://www.ezlan.net/faq.html#transfer

 

[DisgruntledVirus]

Originally posted by: JackMDS
What ever switch is used 100MB/sec. or Giga switch should not make any difference waht so ever for playing on line.

You first line of the trace is from the computer to the Router and it is always 1MS as it should.

In all instances at the second trace you are already out on the Internet.

Unless you played with the Computer's TCP/IP setting and sc*w some thing up, or you have a manged switch that set incorrectly, this should not happen.

Measuring the performance of Network Devices on a LAN via Gaming Server on the Internet is "Silly" at Best.

You want to evaluate your switch and local computers, http://www.ezlan.net/faq.html#transfer

Correct. I realize using games to measure performance is "silly" at best, and agree with you. It's just that's where I noticed it, and I don't know of any tools that allowed me to test local performance (outside of file transfer from PC>PC).

Which is why I turned here. I can troubleshoot managed switches and Cisco/enterprise gear, but an unmanaged switch I'm clueless as to why these results are happening.

I also realize, that online gaming is not a great benchmark and after the router I am on the internet. That said, I don't know how to determine if it's a pc or the router not playing nice with the switch, or if it's something bad in the switch itself.

This is a D-Link DGS-1008D. I'll download that utility and see if I can determine anything else from the results. Thanks for the link.

 

[Madwand1]

Try iperf version 1.7. http://www.noc.ucf.edu/Tools/Iperf/default.htm

server: iperf -s
client: iperf -c server -l 64k -t 15 -i 3 -r

Where server is the name or IP of a remote machine running iperf -s.

E.g. (going across a DGS-2205):

F:\tools\bench\iperf>iperf -c vista-p5q -l 64k -t 15 -i 3 -r
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
------------------------------------------------------------
Client connecting to vista-p5q, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[600] local 192.168.0.189 port 7479 connected with 192.168.0.187 port 5001
[ ID] Interval Transfer Bandwidth
[600] 0.0- 3.0 sec 352 MBytes 984 Mbits/sec
[600] 3.0- 6.0 sec 350 MBytes 979 Mbits/sec
[600] 6.0- 9.0 sec 349 MBytes 977 Mbits/sec
[600] 9.0-12.0 sec 349 MBytes 977 Mbits/sec
[600] 12.0-15.0 sec 350 MBytes 977 Mbits/sec
[600] 0.0-15.0 sec 1.71 GBytes 978 Mbits/sec
[576] local 192.168.0.189 port 5001 connected with 192.168.0.187 port 49570
[ ID] Interval Transfer Bandwidth
[576] 0.0- 3.0 sec 353 MBytes 986 Mbits/sec
[576] 3.0- 6.0 sec 351 MBytes 982 Mbits/sec
[576] 6.0- 9.0 sec 351 MBytes 982 Mbits/sec
[576] 9.0-12.0 sec 351 MBytes 982 Mbits/sec
[576] 0.0-15.0 sec 1.72 GBytes 982 Mbits/sec

 

[DisgruntledVirus]

Okay I used iperf and here are the results:

Using the router (10/100 ports)

C:\Documents and Settings\DV>iperf -c 192.168.1.8 -l 64k -t 15 -i 3 -r
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
------------------------------------------------------------
Client connecting to 192.168.1.8, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[1772] local 192.168.1.3 port 4452 connected with 192.168.1.8 port 5001
[ ID] Interval Transfer Bandwidth
[1772] 0.0- 3.0 sec 34.1 MBytes 95.2 Mbits/sec
[1772] 3.0- 6.0 sec 33.9 MBytes 94.9 Mbits/sec
[1772] 6.0- 9.0 sec 33.9 MBytes 94.7 Mbits/sec
[1772] 9.0-12.0 sec 33.9 MBytes 94.7 Mbits/sec
[1772] 12.0-15.0 sec 33.9 MBytes 94.7 Mbits/sec
[1772] 0.0-15.3 sec 170 MBytes 93.3 Mbits/sec
[1956] local 192.168.1.3 port 5001 connected with 192.168.1.8 port 3225
[ ID] Interval Transfer Bandwidth
[1956] 0.0- 3.0 sec 33.8 MBytes 94.5 Mbits/sec
[1956] 3.0- 6.0 sec 33.8 MBytes 94.4 Mbits/sec
[1956] 6.0- 9.0 sec 33.8 MBytes 94.5 Mbits/sec
[1956] 9.0-12.0 sec 33.8 MBytes 94.4 Mbits/sec
[1956] 12.0-15.0 sec 33.8 MBytes 94.4 Mbits/sec
[1956] 0.0-15.0 sec 169 MBytes 94.4 Mbits/sec

C:\Documents and Settings\DV>

Using the gigabit switch:

C:\Documents and Settings\DV>iperf -c 192.168.1.8 -l 64k -t 15 -i 3 -r
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
------------------------------------------------------------
Client connecting to 192.168.1.8, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[1772] local 192.168.1.3 port 4664 connected with 192.168.1.8 port 5001
[ ID] Interval Transfer Bandwidth
[1772] 0.0- 3.0 sec 832 KBytes 2.27 Mbits/sec
[1772] 3.0- 6.0 sec 1.44 MBytes 4.02 Mbits/sec
[1772] 6.0- 9.0 sec 768 KBytes 2.10 Mbits/sec
[1772] 9.0-12.0 sec 1.25 MBytes 3.50 Mbits/sec
[1772] 12.0-15.0 sec 0.00 Bytes 0.00 bits/sec
[1772] 15.0-18.0 sec 0.00 Bytes 0.00 bits/sec
[1772] 0.0-21.2 sec 4.31 MBytes 1.71 Mbits/sec
[1956] local 192.168.1.3 port 5001 connected with 192.168.1.8 port 3342
[ ID] Interval Transfer Bandwidth
[1956] 0.0- 3.0 sec 1.99 MBytes 5.57 Mbits/sec
[1956] 3.0- 6.0 sec 86.3 MBytes 241 Mbits/sec
[1956] 6.0- 9.0 sec 17.1 MBytes 47.7 Mbits/sec
[1956] 9.0-12.0 sec 26.1 MBytes 72.9 Mbits/sec
[1956] 12.0-15.0 sec 0.00 Bytes 0.00 bits/sec
[1956] 0.0-16.6 sec 132 MBytes 66.5 Mbits/sec

C:\Documents and Settings\DV>

Edit: Just for kicks, I tried everything on different ports to see if it was an individual port issue. Here's the result:

C:\Documents and Settings\DV>iperf -c 192.168.1.8 -l 64k -t 15 -i 3 -r
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
------------------------------------------------------------
Client connecting to 192.168.1.8, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[1772] local 192.168.1.3 port 4740 connected with 192.168.1.8 port 5001
[ ID] Interval Transfer Bandwidth
[1772] 0.0- 3.0 sec 448 KBytes 1.22 Mbits/sec
[1772] 3.0- 6.0 sec 896 KBytes 2.45 Mbits/sec
[1772] 6.0- 9.0 sec 1.50 MBytes 4.19 Mbits/sec
[1772] 9.0-12.0 sec 1.50 MBytes 4.19 Mbits/sec
[1772] 12.0-15.0 sec 2.69 MBytes 7.51 Mbits/sec
[1772] 0.0-15.4 sec 7.06 MBytes 3.86 Mbits/sec
[1956] local 192.168.1.3 port 5001 connected with 192.168.1.8 port 3487
[ ID] Interval Transfer Bandwidth
[1956] 0.0- 3.0 sec 5.26 MBytes 14.7 Mbits/sec
[1956] 3.0- 6.0 sec 8.43 MBytes 23.6 Mbits/sec
[1956] 6.0- 9.0 sec 0.00 Bytes 0.00 bits/sec
[1956] 9.0-12.0 sec 23.4 MBytes 65.5 Mbits/sec
[1956] 12.0-15.0 sec 0.00 Bytes 0.00 bits/sec
[1956] 0.0-16.2 sec 37.3 MBytes 19.3 Mbits/sec

C:\Documents and Settings\DV>

 

[Madwand1]

Try a direct cable connection between the two computers (you'll need to adjust for IPs) -- if that gets good performance, it points to the switch being the problem.

 

[drebo]

You have a duplex mismatch somewhere, no doubt. Or a bad patch cable.

 

[DisgruntledVirus]

Originally posted by: Madwand1
Try a direct cable connection between the two computers (you'll need to adjust for IPs) -- if that gets good performance, it points to the switch being the problem.

Results from that:

C:\Documents and Settings\DV>iperf -c 192.168.1.2 -l 64k -t 15 -i 3 -r
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
------------------------------------------------------------
Client connecting to 192.168.1.2, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[1772] local 192.168.1.1 port 1280 connected with 192.168.1.2 port 5001
[ ID] Interval Transfer Bandwidth
[1772] 0.0- 3.0 sec 258 MBytes 723 Mbits/sec
[1772] 3.0- 6.0 sec 265 MBytes 742 Mbits/sec
[1772] 6.0- 9.0 sec 269 MBytes 751 Mbits/sec
[1772] 9.0-12.0 sec 269 MBytes 752 Mbits/sec
[1772] 12.0-15.0 sec 271 MBytes 757 Mbits/sec
[1772] 0.0-15.0 sec 1.30 GBytes 744 Mbits/sec
[1660] local 192.168.1.1 port 5001 connected with 192.168.1.2 port 4070
[ ID] Interval Transfer Bandwidth
[1660] 0.0- 3.0 sec 335 MBytes 936 Mbits/sec
[1660] 3.0- 6.0 sec 333 MBytes 932 Mbits/sec
[1660] 6.0- 9.0 sec 335 MBytes 938 Mbits/sec
[1660] 9.0-12.0 sec 335 MBytes 936 Mbits/sec
[1660] 12.0-15.0 sec 335 MBytes 937 Mbits/sec
[1660] 0.0-15.0 sec 1.63 GBytes 935 Mbits/sec

C:\Documents and Settings\DV>

So, it's obviously a switch issue (I used the same cables as with the router and switch setups earlier). The previous owner said he had no issues with it, and it gave him full wire speeds (I trust that statement from him, he has 150+ heat and is an active member here).

 

[DisgruntledVirus]

Originally posted by: drebo
You have a duplex mismatch somewhere, no doubt. Or a bad patch cable.

Used completely different patch cables when I switched the ports earlier. So it's not a cable issue.

I have no way of changing the duplex settings on the switch, and based on the other results I think it's fairly safe to assume that it's not a PC side duplex setting. I could be wrong though.

 

[Madwand1]

Try default settings on NICs if you haven't already. Then I suggest returning it to the seller if it doesn't improve significantly. I've tried a few different older DGS-1008D myself (rev C1, C3, C5 IIRC), and never had any trouble with duplex, linking, or (non-jumbo frame) speed.

 

[DisgruntledVirus]

Originally posted by: Madwand1
Try default settings on NICs if you haven't already. Then I suggest returning it to the seller if it doesn't improve significantly. I've tried a few different older DGS-1008D myself (rev C1, C3, C5 IIRC), and never had any trouble with duplex, linking, or (non-jumbo frame) speed.

All settings are default. I rarely mess with the NIC settings, but in this instance they have not been changed (outside of the IP for the previous test). FWIW this is a rev C1.

 

[spidey07]

You have a cabling problem. You cannot use process of elmination with things like this.

That or a duplex mismatch. Your symptoms are classic.

 

[DisgruntledVirus]

Originally posted by: spidey07
You have a cabling problem. You cannot use process of elmination with things like this.

That or a duplex mismatch. Your symptoms are classic.

When I get home tonight I'll try completely different cables (although the cables work fine with the router so I don't see how the cables would be an issue).

If that doesn't work I'll look at all the duplex settings for all the PC's, but since I can't with the switch or router the only possible duplex issue would have to arise from the PC's (or correct me if I'm wrong).

 

[dphantom]

Originally posted by: DisgruntledVirus

Originally posted by: spidey07
You have a cabling problem. You cannot use process of elmination with things like this.

That or a duplex mismatch. Your symptoms are classic.

When I get home tonight I'll try completely different cables (although the cables work fine with the router so I don't see how the cables would be an issue).

If that doesn't work I'll look at all the duplex settings for all the PC's, but since I can't with the switch or router the only possible duplex issue would have to arise from the PC's (or correct me if I'm wrong).

You could also try manually configuring duplex settings on your PC NICs to see if that helps. Otherwise, I am with others here it is a cable/duplex issue somewhere.

 

[Madwand1]

Originally posted by: DisgruntledVirus
Used completely different patch cables when I switched the ports earlier. So it's not a cable issue.

Have you tried factory-terminated cat 5e cables on both ends?

It's unlikely to be a duplex mismatch if you've left the NICs on auto-negotiation as you should. It might be a cabling issue though because a direct link only tests one cable, and 10/100 has lower cabling requirements than gigabit.

 

[seepy83]

Originally posted by: Madwand1
It's unlikely to be a duplex mismatch if you've left the NICs on auto-negotiation as you should.

There could very easily be a driver problem with the NIC, or a firmware problem with the switch, that causes it to auto-negotiate incorrectly.

 

[kevnich2]

Are the cables cat5 or cat5e and home made cables or factory made cables? Either a cabling issue or switch issue I'd say.

 

[DisgruntledVirus]

Originally posted by: Madwand1

Originally posted by: DisgruntledVirus
Used completely different patch cables when I switched the ports earlier. So it's not a cable issue.

Have you tried factory-terminated cat 5e cables on both ends?

It's unlikely to be a duplex mismatch if you've left the NICs on auto-negotiation as you should. It might be a cabling issue though because a direct link only tests one cable, and 10/100 has lower cabling requirements than gigabit.

All cables are factory terminated.

All cables are Cat 5e or higher rated (some of the shorter ones are Cat 6e actually)

I haven't touched the NICs auto-negotiation.

If they were crappy cables though (got them at Microcenter for the long runs) it could cause the issue. When I get home I will wire everything using cat 6e, and completely different cables entirely. That should completely eliminate any possibility of a cable issue. They will also be run differently which will eliminate any possibility of an interference issue.

 

[DisgruntledVirus]

Originally posted by: seepy83

Originally posted by: Madwand1
It's unlikely to be a duplex mismatch if you've left the NICs on auto-negotiation as you should.

There could very easily be a driver problem with the NIC, or a firmware problem with the switch, that causes it to auto-negotiate incorrectly.

The NICs are integrated on the mobos. The one on my desktop is the DFI Lanparty AM2 UT 590 SLI MCP, and the server/HTPC is the Gigabyte GA-MA780G-SH2 AM2+ board. All NICs are left on auto-negotiation.

Anybody know how to ensure the switch isn't having the mismatch?

 

[DisgruntledVirus]

Originally posted by: seepy83

Originally posted by: Madwand1
It's unlikely to be a duplex mismatch if you've left the NICs on auto-negotiation as you should.

There could very easily be a driver problem with the NIC, or a firmware problem with the switch, that causes it to auto-negotiate incorrectly.

The NICs are integrated on the mobos. The one on my desktop is the DFI Lanparty AM2 UT 590 SLI MCP, and the server/HTPC is the Gigabyte GA-MA780G-SH2 AM2+ board. All NICs are left on auto-negotiation.

Anybody know how to ensure the switch isn't having the mismatch?

 

[seepy83]

Originally posted by: DisgruntledVirus

Originally posted by: seepy83

Originally posted by: Madwand1
It's unlikely to be a duplex mismatch if you've left the NICs on auto-negotiation as you should.

There could very easily be a driver problem with the NIC, or a firmware problem with the switch, that causes it to auto-negotiate incorrectly.

The NICs are integrated on the mobos. The one on my desktop is the DFI Lanparty AM2 UT 590 SLI MCP, and the server/HTPC is the Gigabyte GA-MA780G-SH2 AM2+ board. All NICs are left on auto-negotiation.

Anybody know how to ensure the switch isn't having the mismatch?

Does the switch have LEDs for each port that indicate speed/duplex? With an unmanaged switch, you really don't have many options...

 

[RadiclDreamer]

Force 1000/Full and see the results
Then try 100/Full

etc etc

See if it makes any difference

 

[DisgruntledVirus]

Originally posted by: seepy83

Originally posted by: DisgruntledVirus

Originally posted by: seepy83

Originally posted by: Madwand1
It's unlikely to be a duplex mismatch if you've left the NICs on auto-negotiation as you should.

There could very easily be a driver problem with the NIC, or a firmware problem with the switch, that causes it to auto-negotiate incorrectly.

The NICs are integrated on the mobos. The one on my desktop is the DFI Lanparty AM2 UT 590 SLI MCP, and the server/HTPC is the Gigabyte GA-MA780G-SH2 AM2+ board. All NICs are left on auto-negotiation.

Anybody know how to ensure the switch isn't having the mismatch?

Does the switch have LEDs for each port that indicate speed/duplex? With an unmanaged switch, you really don't have many options...

Switch

No, it doesn't have duplex LED's.

Originally posted by: RadiclDreamer
Force 1000/Full and see the results
Then try 100/Full

etc etc

See if it makes any difference

I'll try that as well

 

[JackMDS]

FWIW.

My network (all computers with onboard Giga NIC) was on the switch of a regular Router and yielded about 8MB/sec. local transfer.

I left every thing as is just plugged the computers and the Router to a rock bottom inexpensive Netgear Switch I mange to get about 20MB/sec. LAN transfer.

I left again every thing as is and replaced the D-Link with an Inexpensive Trendnet Giga switch. Now I am mastering 58MB/sec. LAN transfer (that is a real transfer of large files not he short bursts of ipref). :thumbsup:

I don not know whether the specific unit of the Netgear was Bad or that is the performance of all of them.

I basically stop buying Belkin, D-Link, and Netgear. Here and their they might release a good model but most of the time their Hardware quality is random.