PdhineshK,
Your CPU utilization does look a little high, but if I understand correctly that there is a constant delay entering any command into the Nexus CLI, then it is probably not your CPU utilization. There are 2 likely causes, and we'll go over the most likely first.
1) Congestion on the network. I'm not sure how many hops to get to your Nexus switch, or if the Nexus is bogged down itself. Any dedicated SSH traffic (which you probably do want to allocate some bandwidth to) would be shown in the running config as a class map. With the show run command you should see some output that looks like this:
Policy-map LLQ
Class-map MANAGEMENT
Priority 2000 (kbps)
You can also run show policy-map interface gigabitethernet x/x and look for anything related to management or low latency cueing. If you don't have any classmaps I would recommend adding at least one service class to reduce cueing time and see if that solves your problem (these commands are approximate as i don't have a Nexus in front of me to test them exactly):
Nexus(config)#: class-map MANAGEMENT
Nexus(config-cmap)#: match type ssh
Nexus(config)#: policy-map LLQ
Nexus(config-pmap)#: class MANAGEMENT
Nexus(config-pmap-c)#: priority percent 5 (this command will allocate a minimum of 5% of your bandwidth to SSH traffic regardless of other network traffic. It's extremely unlikely you'd ever hit 5% with simple ssh traffic, honestly 1% is pretty slim, so you can choose your own value here or use a kbps measurement)
Nexus(config)#: interface gigabitethernet 0/0 (obviously select your trunk link here)
Nexus(config-if)#:Service-policy output LLQ
Those commands will at least make sure that your SSH traffic doesn't have any obstacles between itself and the way back to your terminal. You may wish to also apply a LLQ service policy in the direction of the trunk link to ensure that all your management traffic gets priority movement across the network.
The other thing I would check right off would be the CoPP status of your device. If you aren't familiar with Control Plane Policing (CoPP) it's essentially just a security policy to keep your CPU from getting overwhelmed or taken down by a DOS attack (by limiting the number of packets that can travel to the CPU). It's unlikely it's interfering, but I would check the policy anyway since your only real latency complaint is with control plane commands (configuration via ssh). Use the following commands to the CoPP and let me know what the output is:
Nexus#: show copp status
Nexus#: show policy-map interface control-plane
I hope some of this information helps you hunt down the problem, and I'll be happy to keep troubleshooting with you once we have some more output. Thanks for posting, good luck!
-Zac
