[c-nsp] High SNMP CPU with SXH. Is SXI any better?

[Phil Mayers]

Jeff Fitzwater wrote:
> 	We are running 12.2SXH2a on sup720-CXL and have been having  
> consistently high (80-100%) CPU on route processor when retrieving  
> either the ARP table or the Bridge-mac table.   No matter what program  
> we use HP NNM or an snmp script, the CPU route process goes from 20%  
> to 90+% with the SNMP process being the top dog when you do a "sho  
> proc cpu sort".

How big is your ARP table? We see high-ish CPU usage with >6k entries, 
even on SXF. Identically configured routers with ~4k entries see much 
lower CPU usage.

There seems to be a non-linear growth in CPU usage for polling (the ARP 
table at least). I do have a TAC case open with Cisco (how bad is it 
going to be with ~10k entries?) but their response has been... poor.

There's some docs on the Cisco website from days of yore, advising 
people to enable CEF, because it provides lexically-sorted data 
structures allowing the ipNetToMedia table to be rendered quickly. 
Obviously this does not apply to 6500s.

I suspect the 6500s don't store the ARP table in OID-lexical order 
internally, and the CPU is having to sort the table every time. I 
suspect the same applies to the dot1dTpFdb table.

> 
> I have looked at every option on list and at CISCO but nothing  
> resolves the problem.   It appears that internally the route processor  
> is doing a lot of crunching to get this table data, specifically the  
> ARP and Bridge Mac table.   I remember something about the format it's  
> in and it had to be converted when retrieved with SNMP.
> 
> 
> Q.   Does anybody know if there is any change with SXI and SNMP queries?
> 
> I also remember reading something about a different way to retrieve  
> this data locally on the router and push it to a host, but cannot find  
> any reference to it now.  Any ideas on this?

There's some support for bulk data gathering

http://www.cisco.com/en/US/docs/ios/12_0s/feature/guide/gdatacol.html

...however if (as I suspect) the