CSR CPU Utilization vs AWS Monitoring

[chun.ning]

We have a CSR deployed in AWS and it is handling about 100-200 Mbps of traffic.

Looking at AWS CloudWatch Monitoring, the EC2 instance hosting the CSR is running at 15-20% (see attached). And looking at the CPU utilization from the CSR console, I see 0%.

CPU utilization for five seconds: 0%/0%; one minute: 0%; five minutes: 0%
 PID Runtime(ms)     Invoked      uSecs   5Sec   1Min   5Min TTY Process 
   1           4          27        148  0.00%  0.00%  0.00%   0 Chunk Manager    
   2       45988     1167506         39  0.00%  0.00%  0.00%   0 Load Meter       
   3          16        2858          5  0.00%  0.00%  0.00%   0 CRYPTO IKMP IPC  
   4           0           1          0  0.00%  0.00%  0.00%   0 Retransmission o 
   5           0           1          0  0.00%  0.00%  0.00%   0 IPC ISSU Dispatc 
   6           6          12        500  0.00%  0.00%  0.00%   0 RF Slave Main Th 
   7           0           1          0  0.00%  0.00%  0.00%   0 EDDRI_MAIN       
   8           0           1          0  0.00%  0.00%  0.00%   0 RO Notify Timers 
   9     1236627      791197       1562  0.00%  0.02%  0.00%   0 Check heaps      

Which one should I trust?

Thanks

- Alan

[lemontree_61089]

Have you been able to find an answer to this question?

[chun.ning]

Yes, but not from Cisco Support.

I was able to derive the real usage data from the peered routers and
confirmed that AWS CloudWatch Monitoring is accurate and CSR's internal
numbers are wrong.

If you are experiencing similar situations, trust your Cloud Service
Provider since they can gather metrics from a lower level (probably
hypervisor) than what Cisco software can see.

... Alan

[lemontree_61089]

Thanks for the feedback Alan.

Will open a case with Cisco to get an official answer from them. Will let you know if I get anything interesting.

Thomas

[pingduck]

What you are seeing from CSR is the CPU utilization of IOS. On the CSR, IOS is only the Control Plane. Most of the heavy lifting is done in the Data Plane. There is another command that can show data plane utilization: "show plat h q a datapath utilization".

[chun.ning]

The CPU utilization from Cloudwatch is about 20%. It is passing about 100 Mbps. The load from your command shows 2%.

show plat h q a datapath utilization
  CPP 0: Subdev 0            5 secs        1 min        5 min       60 min
Input:  Priority (pps)            0            0            0            0
                 (bps)            0            0            0            0
    Non-Priority (pps)        11055        12938         9952         9096
                 (bps)     76791720     93478728     70579240     64994880
           Total (pps)        11055        12938         9952         9096
                 (bps)     76791720     93478728     70579240     64994880
Output: Priority (pps)            0            0            0            0
                 (bps)            0            0            0            0
    Non-Priority (pps)        11054        12938         9952         9096
                 (bps)     78292208     95695304     72121480     66453176
           Total (pps)        11054        12938         9952         9096
                 (bps)     78292208     95695304     72121480     66453176
Processing: Load (pct)            2            2            2            1