What's new

10GbE Testing

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

Butzi

New Around Here
Anyone have experience benchmarking 10GbE switches? I'm having trouble finding a test program/suite that can handle 10GbE bandwidth. Most of them are under-reporting.

TIA.
 
I haven't specifically dealt with benchmarking any actual 10GbE devices but have found iperf to work fairly good for testing raw network throughput. To see how well it could work at 10GbE speeds I loaded up two Windows Server 2008 VMs with each one having a VMNET3 virtual NIC which is supposed to run at 10GbE. From testing this configuration it looks like iperf is capable of saturating a 10GbE connection however a few parallel threads are needed to get there.

Here is the command line I used which was able to saturate the virtual 10 GbE network I setup.

Server:
iperf -s

Client:
iperf -c 169.220.50.100 -w 512k -P4

Notes: This command line only tests in one direction. If you add the -r option at the end it should test one direction and then the other but I could not get it working for me.

Hope it helps.
00Roush
 
Last time I was on one they only gave us 7GBE out of the 10GBE.
 
Last time I was on one they only gave us 7GBE out of the 10GBE.
Something wrong with adapters configuration.
At direct connected (without switch) Windows 2008 Server and Windows 7 PC I have 9.5 - 9.7Gbit with ntttcp test.
AIM (adaptive interrupt management) and jumbo frames was enabled at both adapters (Myricom).
Without AIM I have around 20% performance lost.
 
Something wrong with adapters configuration.
At direct connected (without switch) Windows 2008 Server and Windows 7 PC I have 9.5 - 9.7Gbit with ntttcp test.
AIM (adaptive interrupt management) and jumbo frames was enabled at both adapters (Myricom).
Without AIM I have around 20% performance lost.

This was tested by me on enterprise network domain with over 5,000x Cisco catalyst switches. Of course at the time our backbone was being refreshed under going network refresh still on the new system 10GBE the high was only 7GB.

But you tested a different way than I did.. So your results will vary.
 

Latest threads

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top