[rtg] Fwd: Results with RTG

[Brandon Ewing]

On Mon, Mar 09, 2009 at 04:32:01PM -0400, John Von Essen wrote:
>
> I haven't seen any real world stats on the super high-end Cisco gear, but I 
> know from experience Sup2 and Sup32 on 6500/7600 really start to break down 
> once you hit like a few hundred Mbps. The Netflow memory cache gets 
> reached, and you have to start sampling.
>

IIRC, it's not the collection that kills a Sup, it's the exporting.  With
6500s, it's a dance between TCAM (storing flows til expiry) and CPU
(exporting expired flows).  "Sampling" on a 6500 is sampling the flow table
at expiration time (all flows are still tracked in TCAM according to
flowmask) -- it's GSR's that pick random packets to build flows when
sampling.

Are we off-topic yet?

-- 
Brandon Ewing                                        (nicotine at warningg.com)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.grdata.com/pipermail/rtg/attachments/20090309/1d6afc94/attachment.pgp>