07-04-2010 09:10 PM - edited 03-04-2019 08:57 AM
Hi All,
Hope that all are doing well.
This post is targetting to get a confirmation about CPU utilization strategy of Cisco 7206VXR (NPE-G2) platform.
CPU utilization report of our 7206 router is as follows.
***************************************************************************************
Show Process CPU sorted 5min | exclude 0.00%
CPU utilization for five seconds: 69%/67%; one minute: 69%; five minutes: 69%
PID Runtime(ms) Invoked uSecs 5Sec 1Min 5Min TTY Process
77 6434145241506095648 427 1.63% 1.64% 1.65% 0 IP Input
57 1990322054277071 0 0.16% 0.18% 0.16% 0 T3E3 EC IPC POLL
46 95016 53336837 1 0.08% 0.06% 0.08% 0 Per-Second Jobs
***************************************************************************************
During peak business hours, CPU utilization is going up to 70% out of which 68% is interrupt utilization. There are 5 POS interfaces (155M) and each POS is carrying around 148M incoming traffic and 50M outgoing traffic. One Gig interface is carrying 824M outgoing traffic and 265M incoming traffic. All interfaces are processing packets using CEF technology.
There are no misbehaving processes / logs observed from the router logs.
Show Version depicts the following. Bandwidth points have been reached its limit.
*************************************************************
System image file is "disk2:c7200p-advipservicesk9-mz.124-11.T2.bin"
PCI bus mb1 (Slots 1, 3 and 5) has a capacity of 600 bandwidth points.
Current configuration on bus mb1 has a total of 600 bandwidth points.
This configuration is within the PCI bus capacity and is supported.
PCI bus mb2 (Slots 2, 4 and 6) has a capacity of 600 bandwidth points.
Current configuration on bus mb2 has a total of 600 bandwidth points.
This configuration is within the PCI bus capacity and is supported.
*************************************************************
I am planning to fine tune the router configuration and knock off unneeded BGP peership / ACL / Static routes and route-maps. As per my understanding, under these conditions 70% CPU utilization is normal.
It would be a great help, if anyone can confirm the following.
-> Under these conditions, Is CPU utilization of 70%/68% OK ?
-> Is there any technical document which states the CPU baseline for NPE-G2 platform?
Thanks in advance....
Best Regards,
Manjesh Michael
07-05-2010 12:05 AM
If you are using acl more than 50 lines then better to use complied acls. Are your BGP and IGP sessions are stable?
regards
Shivlu Jain
07-05-2010 01:35 AM
Hello,
Taking in count the WAN links and their utilization in your router and the NPE-G2 routing capabilities (2M pps/1024Mbps) I would say having spikes of 70% during the peak hours is not an issue and they may be caused by the high traffic rate.
Just to be sure this spikes are not overloading the router forwarding capacity I would check any input or output drops on the POS interfaces. If they are not present or they represent <1% of the total traffic, the router would be working as expected.
Hope this helps.
07-05-2010 04:50 AM
Hi,
Thanks Shivlu and Jorge.... for your timely suggestions.
-> Access-list compiled is already in place. I don't see any protocol fluctuations (BGP / OSPF).
-> Presenlty there are no I/P or O/P drops on POS / Gig interfaces.
I feel the CPU hike is purely related to the traffic rate. During non-business hours, CPU utilization is going down from 70% to 25-30%.
If any technical document / URL is availabe for NPE-G2 performance strategy, please share it.
Thanks in Advance,
Best Regards,
Manjesh Michael
07-06-2010 01:04 PM
Hello Manjesh,
>> During peak business hours, CPU utilization is going up to 70% out of which 68% is interrupt utilization. There are 5 POS interfaces (155M) and each POS is carrying around 148M incoming traffic and 50M outgoing traffic. One Gig interface is carrying 824M outgoing traffic and 265M incoming traffic. All interfaces are processing packets using CEF technology.
There are no misbehaving processes / logs observed from the router logs.
NPE-G2 should be able to move 2,000,000 packet per second I would say this box is too loaded, be also aware that several colleagues have reported inferior performances when using WAN port adapters instead of built in GE ports
even if CEF is enabled the router works only in software on main cpu.
Hope to help
Giuseppe
07-23-2010 08:46 PM
Hi Giuseppe,
Thanks for the reply.
We have done a detailed analysis of router configuration part and accomplished the following fine tuning activities.
-> Fine tune BGP configuration and remove unwanted BGP peership and network statements.
-> Removed unneeded PBR applied on the interface
-> Reomved unneeded static routes, loggind entries and SNMP traps commands
CPU utilization has come down slightly and found that CPU statistics is stable.
Best Regards,
Manjesh Michael
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