[c-nsp] Maximum traffic on Gigabit Ethernet

[Nick Hilliard]

On 03/10/2011 20:57, Paul Cosgrove wrote:
> I think you may have an error there.  Your calculations suggest that
> traffic in on a 1Gbps interface will fill a 1.3M Byte receive buffer in a
> little under a 1/10 of a second.  While the preamble and inter frame gap
> will slow the buffer fill rate (the degree depending on the packet size),
> that seems very slow.  

yeah, I was sloppy with language there, and ignored ipg.  I also assumed 8
bytes per bit, maximum buffers assigned to tx data (in fact it's 1.2M tx
and 166k rx on a 6748-ge blade), flow control disabled and I wasn't clear
about the exact direction of data, and so on...

The point was that if you allow port contention (i.e. by enabling flow
control) and if you only look at 5 minute port counters, you are only
guaranteed not to drop packets at ridiculously low apparent data transfer
rates.  5 minutes is an eternity in the life of a switch.

Bah, I should have known not to pick nits like this in public!  It never
turns out well. :-(

Nick