Cisco 4500 performance issues

[zhangziyi]

Hi, being more of a MS person, I'm not too sure on this, but would be a great help if someone could provide some help.

We have a 100Mb/s link out of our building. This is presented as a fibre link LAN extension, into a convertor (NTU) which converts to UTP.

This UTP is connected via a patch lead into one of the ports on the Cisco 4500.

After tests, we're getting 4Mb/s down and 50Mb/s up over that link from a PC on our LAN (through the 4500)

Plugging a laptop directly into the NTU using a x-over cable (and therby bypassing the switch) acheives 70Mb/s both ways.

Must be the switch then, we're tried different devices and O/S's Linux, MacOS and obviosuly Windows XP.

Any idea, could this be a Cisco CDP, STP, Routing issue?

Thanks in advance!

 

[ADB100]

Most likely a duplex mismatch between the NTU and the switch, unless you are plugging the LES straight into your LAN and it is real busy?

Andy

 

[zhangziyi]

Thanks for the reply, sounds good. The port is set to 100 Full, no cdp enable. Apologies is this sounds a little silly, but what would a LES be? Cheers Mike.

 

[vipergg]

If its slow and the switchport is hardcoded to 100/full then it is most likely that the the NTU is setup for "auto for speed and duplex and you have a speed/duplex mismatch which would cause the ntu to default to 100/half and cause all kinds of collisions and errors , try setting the switchport to auto for speed and duplex and see what happens.

 

[zhangziyi]

Excellent advice, I'll raise a change control and try on Monday. Thanks very much

 

[ADB100]

LES - LAN Extension Service. WAN Services provided over Ethernet - 10, 100 or 1000Mbps

You say you plugged a PC directly in and got 70Mbps - did you hard-code the PC's NIC to 100/Full? If not then I think we are on the right tracks with a Duplex mismatch.

HTH

Andy

 

[zhangziyi]

Almost sure that the NIC was set to Auto, we didn't configure it.

Nice one, thanks everyone for the help

Will let you know how it goes on Monday!

Cheers,

Mike

 

[zhangziyi]

Ahh LES, so that's what it is, thanks.

Because it's providing WAN services over ethernet, would it use any routing?

Could it be a routing or switching issue?

Cheers Mike

 

[ADB100]

would it use any routing?

That depends on a few things - is it a point-to-point link between offices or is it a link into a service provider for Internet or MPLS etc?

Normally LES circuits are point-to-point links (i.e. /30 subnets) and use routing - either statically or with a routing protocol.

HTH

Andy

 

[zhangziyi]

Hi Andy,

Thanks for your reply, yes you're right it is a point to point link to our data centre and then from the data centre out to the Internet.

Cheers,

Mike

 

[zhangziyi]

Hi, I set the port speed and duplex to auto but hasn't imporved the performance.

Any ideas?

 

[zhangziyi]

Hi,well auto didn't work, here's the output of the show interfaces command. The errors are when I tried to set to 10/half.

Anyway, any ideas?

Cheers

GigabitEthernet8/18 is up, line protocol is up (connected)
Hardware is Gigabit Ethernet Port, address is 0007.ebe4.c129 (bia 0007.ebe4.c129)
MTU 1500 bytes, BW 100000 Kbit, DLY 10 usec,
reliability 255/255, txload 3/255, rxload 1/255
Encapsulation ARPA, loopback not set
Keepalive set (10 sec)
Full-duplex, 100Mb/s, link type is auto, media type is 10/100/1000-TX
input flow-control is off, output flow-control is off
ARP type: ARPA, ARP Timeout 04:00:00
Last input 00:00:01, output never, output hang never
Last clearing of "show interface" counters never
Input queue: 0/2000/0/0 (size/max/drops/flushes); Total output drops: 0
Queueing strategy: fifo
Output queue: 0/40 (size/max)
5 minute input rate 515000 bits/sec, 273 packets/sec
5 minute output rate 1248000 bits/sec, 256 packets/sec
942098 packets input, 503993598 bytes, 0 no buffer
Received 8173 broadcasts (192 multicast)
0 runts, 0 giants, 0 throttles
0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored
0 input packets with dribble condition detected
995322 packets output, 754214489 bytes, 0 underruns
785 output errors, 157 collisions, 0 interface resets
0 babbles, 472 late collision, 0 deferred
0 lost carrier, 0 no carrier
0 output buffer failures, 0 output buffers swapped out

 

[zhangziyi]

In another port, similar results...

FastEthernet7/43 is up, line protocol is up (connected)
Hardware is Fast Ethernet Port, address is 0012.7f6e.cb4a (bia 0012.7f6e.cb4a)
Description: 100mb link to Telewest (white patch 58)
MTU 1500 bytes, BW 100000 Kbit, DLY 100 usec,
reliability 255/255, txload 2/255, rxload 1/255
Encapsulation ARPA, loopback not set
Keepalive set (10 sec)
Full-duplex, 100Mb/s, link type is auto, media type is 10/100BaseTX
input flow-control is unsupported output flow-control is unsupported
ARP type: ARPA, ARP Timeout 04:00:00
Last input 00:00:29, output never, output hang never
Last clearing of "show interface" counters never
Input queue: 0/2000/0/0 (size/max/drops/flushes); Total output drops: 0
Queueing strategy: fifo
Output queue: 0/40 (size/max)
5 minute input rate 491000 bits/sec, 280 packets/sec
5 minute output rate 1101000 bits/sec, 267 packets/sec
6306450403 packets input, 2908852863732 bytes, 0 no buffer
Received 540504464 broadcasts (1179689 multicast)
0 runts, 0 giants, 0 throttles
1 input errors, 1 CRC, 0 frame, 0 overrun, 0 ignored
0 input packets with dribble condition detected
6011669986 packets output, 2593876363283 bytes, 0 underruns
0 output errors, 0 collisions, 0 interface resets
0 babbles, 0 late collision, 0 deferred
1 lost carrier, 0 no carrier
0 output buffer failures, 0 output buffers swapped out

 

[ADB100]

785 output errors, 157 collisions, 0 interface resets
0 babbles, 472 late collision, 0

The interface says 'Full-duplex, 100Mb/s' - you can't have collisions on a Full-Duplex link..... I suggest you post some parts of the config, I'd probably be looking at IOS bugs as well.

Andy

 

[vipergg]

The counters have never been cleared so you can get a good reading of what is going on . Clear the interface counters and watch before doing anything else . It looks like it auto negotiated correctly at 100/full , if it didn't negotiate right it would have defaulted to 100/half , looks right to me . If you clear the counters and don't see any errors then the problem is not the link and you would need to look elsewhere .

 

[jpcomeau]

Are the blades in the 4500 PoE? I have had some issues with servers on a PoE blade behaving oddly until I shut down the power on the port.

 

[zhangziyi]

Problem was due to duplex and speed mismatch on the other end of the LES. Thanks very much for the help!

Cheers.