cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2023
Views
4
Helpful
7
Replies

TTCP - throughput question

Abertay
Level 1
Level 1

Hi,

I've been testing some of our links with ttcp to make sure everything is ok and I don't seem to be getting much more than 350KB/s even testing between two stacked switches.

The highest i've managed to achieve is 1250KB/s between two brand new 3560's directly attached to each other.

Sample output ::

F:\utils\cisco\bandwidth>ttcpw -r -s ttcp-r: buflen=8192, nbuf=2048, align=16384

/0, port=5001 tcp ttcp-r: socket

ttcp-r: buflen=8192, nbuf=2048, align=16384/0, port=5001 tcp

ttcp-r: socket

ttcp-r: accept from 192.x.x.x

ttcp-r: 16777216 bytes in 51.13 seconds = 320.47 KB/sec +++

ttcp-r: 2066 I/O calls, msec/call = 25.34, calls/sec = 40.41

ttcp-r:

thanks.

1 Accepted Solution

Accepted Solutions

Agree with Rick and Joseph.

If you want to test for example Access lists, or something in that nature than you can test from switch to switch.

But if you want to test maximum throughput of the link, it really need to be done PC to PC.

Also, try different tools.

Like "iperf","hping", or FTP server-client transfers.

iperf is really easy one.

With HPING you have to monitor bandwidth load.

FTP will give you real life results, but is limited by PC bus speed/HDD performance etc.

Please rate all helpful posts

View solution in original post

7 Replies 7

jwdoherty
Level 1
Level 1

What's the source and destination of the TTCP tests? If the target is one of the switches, keep in mind the switch is optimized for traffic through the box not to it.

i've been trying doing switch->switch tests is it more accurate to do PC->PC tests through the switches instead?

Matt

I think that the answer depends a bit on what you are really trying to measure. Especially if you are interested in how the switch passes network traffic then it would be best to test PC to PC. If you are trying to measure something in between the switches it probably matters less, though even in that situation it might be better to have a PC as the source and as the destination.

HTH

Rick

HTH

Rick

I've done further testing pc->pc and it is far more accurate for what i'm looking for. All the figures tested so far seem in keeping with the link speed.

I've been using PCATTCP. great little tool.

Matt

I am glad that you got this resolved. And thank you for updating the thread to indicate your progress in resolving this. It makes the forum more useful when people can read about an issue and can read about the resolution of the issue. Perhaps you could use the rating system to indicate that it was resolved so people would recognize its status?

HTH

Rick

HTH

Rick

Just thought i'd add the link for that tool.

http://www.pcausa.com/Utilities/pcattcp.htm

Nice and simple usage too.

on the reciever run :

pcattcp - r

on the sender run :

pcattcp -t Ip-address-of-reciever

Agree with Rick and Joseph.

If you want to test for example Access lists, or something in that nature than you can test from switch to switch.

But if you want to test maximum throughput of the link, it really need to be done PC to PC.

Also, try different tools.

Like "iperf","hping", or FTP server-client transfers.

iperf is really easy one.

With HPING you have to monitor bandwidth load.

FTP will give you real life results, but is limited by PC bus speed/HDD performance etc.

Please rate all helpful posts

Review Cisco Networking products for a $25 gift card