[c-nsp] Nexus 7000

[Aamer Akhter (aakhter)]

There are some techniques that one can use to scale something like this. One could use a UDP loadbalancer to a complex for collectors to distribute the load. 

The cisco NFC does support this model with hiearchrial collectors, as other collector implementations do as well. The one thing you have to look out for is wheter the individual collector is doing some kind of aggregation before sending the data to a higher hierarchy. In such a case you could wind up with misweighted data. Also keep in mind the licensing models for some of the collector software (cisco nfc is per collector install, others are per router port, or per routing device).

Cisco NFC can run on a variety of machines (an actual appliance is not sold, just the software), so a single box's scale could be quite high. I don't have the numbers on a reference system though.

Regards,

-- 
Aamer Akhter / aa at cisco.com
Ent & Commercial Systems, cisco Systems

> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net [mailto:cisco-nsp-
> bounces at puck.nether.net] On Behalf Of James Humphris
> Sent: Tuesday, January 29, 2008 8:00 AM
> To: Lincoln Dale (ltd); cisco-nsp at puck.nether.net
> Subject: Re: [c-nsp] Nexus 7000
> 
> Lincon,
> 
> Just on the netflow point, whilst it's all very well being able to
> generate a ton of netflow data export records, but has the Cisco
> Netflow
> Collector been sufficiently scaled to deal with the sort of volumes of
> traffic generated by these "next generation" platforms?
> 
> What type of hardware platform would I need to use as a collector if I
> enabled netflow (or even sampled netflow) on a number of these devices
> in my data centre!?!
> 
> James.
> 
> 
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Lincoln Dale
> (ltd)
> Sent: 29 January 2008 12:30
> To: Ray Burkholder; cisco-nsp at puck.nether.net
> Subject: Re: [c-nsp] Nexus 7000
> 
> > Does it do netflow output?
> 
> yes, NetFlow export v5 and v9 (flexible netflow) export are supported.
> 
> h/w table of 512K netflow entries shared between ingress & egress on
> each forwarding engine (per I/O module).
> 
> > Or sampled netflow?
> 
> yes, (true) sampled netflow is supported in h/w too.
> 
> > I suppose one would need a
> > small network just to handle the netflow output of a fully traffic'd
> > switch.
> 
> quite possibly!
> 
> in the initial software release the export is handled by the Supervisor
> control-plane. in a maintenance release soon after, we're adding
> distributed netflow export such that its exported from the control-
> plane
> local on the I/O module.
> even with the centralized model, its forseeable that there could be
> substancial export rate!
> 
> 
> cheers,
> 
> lincoln.
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/