RE: Q-in-Q BPDU

Is this a lab or a real-world design?

-----Original Message-----
From: nobody_at_groupstudy.com [mailto:nobody_at_groupstudy.com] On Behalf Of Richard Ginto
Sent: Wednesday, January 27, 2010 1:25 PM
To: ccielab_at_groupstudy.com
Subject: Q-in-Q BPDU

I'm running a Q-in-Q tunnel using 3750s and "customer" switches at the edge
like so:

customer switch #1 <--Dot1Q access--> SW1 <--trunk--> SW2 <--trunk--> SW3
<--Dot1Q access--> customer switch #2

Using shared-round-robin I have the carrier switches setup to place the
customer's EF traffic into the priority queue and I police that traffic
inbound to 5Mbps. I then have the remaining 3 queues weighted so as AF31
goes to a queue with a weighting of 60, AF21 goes into a queue with a
weighting of 40, and the last queue is a weight of 1 which is for BE
traffic.

The problem I see is when I saturate this link during testing with 5Mbps of
EF and 100+ Mbps of AF31 and AF21 traffic, I drop BPDUs between my customer
switches and spanning-tree re-converges.

Any suggestions on how I can get my BPDUs prioritized through this so I
don't drop them during congestion? Though I am doing everything on layer 3
markings, I re-mapped CoS so that Cos 5 is placed into the priority queue
and Cos 6 & 7 are placed into the queue with a weighting of 60.

Thanks all

Blogs and organic groups at http://www.ccie.net
Received on Wed Jan 27 2010 - 16:50:42 ART