[183442] in North American Network Operators' Group
Re: NetFlow - path from Routers to Collector
daemon@ATHENA.MIT.EDU (Roland Dobbins)
Wed Sep 2 05:42:31 2015
X-Original-To: nanog@nanog.org
From: "Roland Dobbins" <rdobbins@arbor.net>
To: nanog@nanog.org
Date: Wed, 02 Sep 2015 16:40:37 +0700
In-Reply-To: <20150902055057.416E3550C6B46@freedman.net>
Errors-To: nanog-bounces@nanog.org
On 2 Sep 2015, at 12:50, Avi Freedman wrote:
> + optimal tweaks include local flow proxy that can also rate
> limit / re-sample, and send topk talkers over 'true' OOB.
Yes, a lot of distributed flow collection/analysis architectures are
deployed this way, doing more than top-talkers, even. The key is that
traffic between the elements of those types of deployments *must* not be
disrupted; true OOB is preferred, but VLAN/VRF is still the norm, there.
-----------------------------------
Roland Dobbins <rdobbins@arbor.net>
