03-14-2011 08:20 PM - edited 03-06-2019 04:05 PM
I am in the process of installing a 3750x (IOS 12.2 (53r) SE2 IP Base) Cisco Catalyst switch in a new network of just 2 PC's (2 hosts, OS windows7 64Bits). I have enabled SVI interfaces with the both hosts installed in 2 different network segments. We then start connectivity test.
The response time for the PING command between both hosts remain below 1 millisecond, whereas the response time between the hosts and their correspondent SVI interface is variable, and at all time is higher than 1 millisecond, sometimes it reaches 17 milliseconds. (Note that the switch CPU usage is only 8% at the time of testing)
We have performed this same connectivity test changing the 3750x switches and in two different locations obtaining the same results.
Can anybody explain if this normal, why the response times to the SVI interface is higher than 1 millisecond? Is this normal?
Thank's
03-18-2011 06:31 PM
It is normal. The pings are being processed in the control plane, not in asic. Additionally, icmp within the control plane has very low priority so other processes will get handled first. Its not enough for a significant spike in CPU but it is enough to add the latency you're experiencing. I ran into the same thing on a pristine 6509 VSS switch in a green-field datacenter buildout. This was the explanation Cisco TAC gave me.
As long as your latency is acceptable when you ping through the device, and the CPU load is reasonable, you're good to go.
05-07-2013 11:07 PM
Hi ajnallicruz,
Do you have the documentation from TAC on why ICMP has high response time when being processed by the Control-plane?
Thanks,
Gene
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