1. Home
  2. Cisco Systems
  3. Troughput issue

Troughput issue

Hi,

We have a network provider that has linked our 3 offices with LANlink. Office A (cisco 4503) is connected to the provider with 1 Gbits fiber optic full duplex Office B (cisco 3750) is connected to the provider with 100 Mbits full duplex Office C (cisco 3750) is connected to the provider with 10 Mbits full duplex

These lines are still on test so they are not in production yet. When I try to do ftp between office A and office B, I got 1 Mbyte/s. Normally I should get around 10 Mbytes/s. I complained to the provider, the technician told to use NetIO command line software. This software do test with packet of 1K, 2 K, 4K, 16K and 32K. So I put on one end the tool on listening mode and on the other end it send the data. Indeed the testing gave a throughput of 10 Mbytes/s (100 Mbits line).

I was not convinced as the packet are too small, I used then another command line software NetCPS. This tool work in the same manner but it send a file of 100 Mbytes. Here the troughput was the same as the ftp: 1 Mbyte/s. It took 73 seconds to send 100 Mbyte file.

The Cisco switches are new and we just use static routes. We do not have any QoS.

I would appreciate if you can give some help on this

Thanks

Reply to
Slimo
Loading thread data ...

sounds like there may be a duplex mismatch.

the classic mistake is to set 1 end as full duplex and expect the other using "auto" to match it - but it would give half duplex since a fixed config port doesnt reply to negotiation.

check the cisco interface info on each port to see if they are running half duplex.

if the carrier has the problem all you can do is get them to check the port settings, or look for a high inbound error rate on the cisco ports during the test.

Also - the same thing could affect your measurement devices.

Reply to
stephen

Thanks. I checked the duplex and both sides of the servers are 100 Mbits / full duplex. I got the following error status for the interface where the fiber optic cable is connected:

switch1>sh int gig1/1 counters errors

Port CrcAlign-Err Dropped-Bad-Pkts Collisions Symbol-Err Gi1/1 16 0 0

234118815

Port Undersize Oversize Fragments Jabbers Gi1/1 22 0 0 1

Port Single-Col Multi-Col Late-Col Excess-Col Gi1/1 0 0 0 0

Port Deferred-Col False-Car Carri-Sen Sequence-Err Gi1/1 0 0 0 765

Reply to
Slimo

just to make sure i was clear - it isnt whether the servers are set to

100/full that is the issue, but whether the switch ports they connect to are consistent for each physical link.

so server 100/full (current status, not configured state) same as connected cisco switch port 100 / full (again measured not just set up).

I got the following error status for the interface where the

this one doesnt look good - if it is still going up you need to check the fibres, patch leads and so on.

Reply to
stephen

I set both switches port and nic to 100 full duplex. What is stange is that doing test with small packet size 1k, 2k, 16K and 64k give a throughput of

10 Mbytes/s. If I try to do a test with a big file, the throughtput drop to 1.5 Mbytes/s.

Reply to
Slimo

it might be down to disk performance on the end stations or similar.

try a network test application instead - we used ttcp at work a while back.

this is a good place to start:

formatting link

Reply to
stephen

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.