High CPU Utilization on Cisco SGE-2000

[dc-csa-blr]

Dear Experts,

We are using Two Cisco SGE-2000 in our network. but we are facing CPU utilization very high upto 95 to 98 percent.

we are unable to see in details why it's happend like in cisco RTR or Catalyst " sh proc cpu" there is see in details by which service it going high so then we can identity very easly but in Cisco L3 SGe-2000 only showing Percent not in details. so how can i find out.

We are getting on SGE-2000 this type of errors "2147480831 2012-Jul-06 13:14:45 Warning %IPFFT-W-SFFTREDYELLOW: IP SFFT"

how can solve this and why it's happend.

Please anyone help me for the same , i appreciate for this.

Thanks in ADV,

[Tom Watts · Accepted solution]

Hi DC, SFFT errors usually indicate too much IP traffic.

-Tom

[dc-csa-blr]

D/Tom,

Thanks for your reply . you are right there is more than 400 users connected with this SW so how can i resolve this issue.

one more think should i go with this SW when connected more than 400 users and also we have configured with Static route and we have facing CPU utilization very high continue 90 to 99 percent. and how can i show in details which service is taking high percent of CPU utilization cause i am not able to show in details in this SW only percentage are showing.

This SW is reliable at this level ?? or not and i want to know how many BW perfonance capacity of this SW currentely we are using 100 mb P2P.

one SW at our CO (SGE-2000) with 150 users and conneted 100 mb P2P with same SGE-2000 at our HO with more than 400 usrs.

there is no any issue at our CO with the same SW.

Thanks once again!!!

[Tom Watts · Accepted solution]

Hi DC, small business products are generally geared for networks 100 users and less or service as edge devices to larger networks in the access layers.

The performance should be okay. But, if you start dropping traffic or see a performance degradation it is because the table over flow stopped forwarding to destination IP address due to insufficient resources because at this point, you are software switching most of the network traffic.

If it is at all possible, and you must maintain 400 users, you should attempt to lower the tcam utilization such as QoS, ACL, vlan rate limiting,  and mac based vlan. Also, if you can use the switches in layer 2 and allow the routing to occur at a more true core layer, you will have better results.

-Tom

[dc-csa-blr]

Dear Tom,

At present already we are configured QoS for Voice/Video high priority over data match with ACL and also multiple VLAN for Voice/Video (Vlan10) and for Data (Vlan1).

But TCAM utilization very low upto 3 to 4 % only.

So what you have to suggest which SW i hve to use at current Scenerio and effecting which service for high CPU utilization in this SW so how can i achive to see cpu utilization in details.

Thanks.

[Tom Watts · Accepted solution]

Hi DC,

The switch doesn't have particular or granular diagnostic to see where it works too much. The logging entries are the indication, which has been directed to IP traffic due to SFFT errors.

Although your TCAM utilization is low, all of the QoS and ACL will subject the ingress traffic. With 400 IP traversing the switch, this will increase the CPU and memory usage. In turn, processing the IP traffic generate such an error.  The most true answer I could offer is to reduce the amount of IP traffic.

But again, if you are not seeing a degradation in network performance, the error may be ignored as the software switching is handling it.

-Tom