[c-nsp] 3560 mtu miss-match causing output drops
Dale Shaw
dale.shaw+cisco-nsp at gmail.com
Mon Mar 15 05:09:35 EDT 2010
This seems at least vaguely relevant at this point in the thread:
On Mon, Mar 15, 2010 at 7:55 PM, Peter Rathlev <peter at rathlev.dk> wrote:
> I've never found out exactly how large the egress buffers are, but I've
> heard it's around 64 packets (* 1500 bytes ~ 100 KB). I that case a
> sustained gigabit feed to a 100 Mbps port (i.e. 900 Mbps excess) can be
> buffered for ~ 1 ms before the switch starts to drop. (Unlees my quick
> calculations are wrong of course.)
Pavel Bykov <slidersv at gmail.com> recently posted a link (on the
GroupStudy CCIE list) to a catalyst QoS calculator he created:
See: http://bitsonwire.com/?p=51
"Ever wanted to configure queue depth on your Sasquatch family switch
using bytes instead of relative percentages? Whenever you configure
QoS on Catalyst 2960, 2960G, 3560, 3560G, 3750, 3750G or 3750ME you
can use the following document for byte calculation, cross check and
reference [...]"
Cheers,
Dale
More information about the cisco-nsp
mailing list