"Prokott, Joe" wrote:
>
> I know this question has been asked before, but I couldn't find info. in my
> archives.
>
> What causes and what is the resolution to high router CPU due to netmon
> polling a Cisco(7505) router?
The extraordinarily inefficient way in which Cisco encapsulates
and dumps the Route Table and Arp Cache during the daily config
poll. Can also be somewhat IOS-dependent...
> Is it the number of routing entries netmon is
> getting from the router?
Yes, and the Arp Cache. Both can be constrained at the router,
depending on your IOS. The Route entries can be constained within
netview in SNMP conf.
>Is the resolution just to configure netmon so it
> gets fewer router entries from the router?
We've found this to be adequate. The facility exists within
the IOS but we haven't had to get that drastic. Yet...
> What are the implications if
> netmon gets fewer routing table entries (i.e., what is NV not able to do
> from lack of this data?)?
Very little, if any. Depends on whether or not you use auto-
discovery. We don't, so have no problems constraining route
entries to 5. I can envision it being a problem if you are
using the route entries for populating root-cause lookups or
some such, but we're (just) static enough that it hasn't yet
been a problem.
> Does NV version 6.x query the routers differently
> so this does not occur?
>
Haven't had v6 up and running long enough to determine this, Joe.
But what I've read and seen so far indicates that there is no
fundamental difference in the actual polling. I could likely
be wrong on this and I'm sure James will speak up if so...
Dave Dimond
Allina Health System
|