09-24-2010 12:53 PM - edited 03-06-2019 01:10 PM
Hi all,
I have a pre-deployment test bed have a 6509 720 sup engine system with a 16 port 10G, and a ASA, a 4900M and VeEX testers. With 2 data streams, 10G traffic generated from tester, the 10G card in 6509 system only can provide about 50% util bandwidth. Also I can not find where those frames were got dropped.
More detail: ten gi port 3 input X frames, but ten gi port 1 output will be much less (both trunk ports). This only occurs when traffic load higher than 5G traffic.
Is any one know any 10G performance test report?
Also does where we plug the 10G card make diff?
Any info would be greatly appreciated.
Thanks.
gy
Solved! Go to Solution.
09-24-2010 01:01 PM
Hello Gongyuan,
WS-6716 has a 4:1 oversubscription and 4 tengiga ports are served by a single ASIC, only one port of the group of 4 ports can be line rate. So if in your tests you were using ports served by same ASIC your results my be explained by this fact.
see
16-Port 10 Gigabit Ethernet Copper Module
09-24-2010 01:01 PM
Hello Gongyuan,
WS-6716 has a 4:1 oversubscription and 4 tengiga ports are served by a single ASIC, only one port of the group of 4 ports can be line rate. So if in your tests you were using ports served by same ASIC your results my be explained by this fact.
see
16-Port 10 Gigabit Ethernet Copper Module
09-24-2010 01:24 PM
Thank you so much for your quick response.
We have 16-Port 10 Gigabit Ethernet Fiber Module.
According to the spec, if I pick two ports in diff port group, then it will reach full rate (I assume port 1-4 in one group, such as), such as I pick up port 1 and 5 (or 7, etc.).
If we use 4-Port 10 Gigabit Ethernet Fiber Module, that should give us full bandwidth.
Thanks again for your great help.
gy
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide