[c-nsp] IP SLAs - throughput measurement

[Bill Nash]

He's talking about dueling gateways. With both boxes sending traffic to 
each other, all packets sent into the network are routed to each other in 
a cyclic loop, each packet living until its TTL is decremented to 0. You 
just keep feeding data into it until the volume gets to the level you 
want, and monitor for errors. Stop sending new data, and the load will 
slowly drop as TTLs expire.

- billn

On Mon, 18 Dec 2006, Adam Greene wrote:

> Ras,
> 
> Thanks for the advice. I'm going to show my ignorance here. I'm not quite 
> sure how that will multiply the traffic load by ~255. I assume you are not 
> saying that I could generate a single ~3.92Mbps traffic stream from PC to 
> PC, and the fact that the switches are routing to each other will cause them 
> to route the traffic on all 255 IP's in the /24, thus multiplying the 
> inter-switch traffic to 1Gbps. There must be something I'm not getting here.
> 
> Would you mind elaborating a bit more for me? I'd like to understand your 
> method better, if possible.
> 
> Thanks,
> Adam
> 
> ----- Original Message ----- 
> From: "Jee Kay" <jeekay at gmail.com>
> To: "Adam Greene" <maillist at webjogger.net>
> Cc: <cisco-nsp at puck.nether.net>
> Sent: Monday, December 18, 2006 11:49 AM
> Subject: Re: [c-nsp] IP SLAs - throughput measurement
> 
> 
> > My usual method of this is mgen and L3 on the switches.
> >
> > Basically, bring up L3 on the switches and set the default routes
> > (assuming they're not already doing L3.. otherwise chose some
> > 192.0.2/24 address space) to point at each other. This nicely
> > multiplies your traffic load by ~255 (assuming default TTLs).
> >
> > Then point mgen (or iperf if you like) at the switch L3 and watch the
> > error counters :) If they're all zero (no overruns, no drops) then
> > you're good.
> >
> > Ras
> > 
> 
> 
> 
> 
> 
> _______________________________________________
> 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/
>