[ednog] Anyone deploying MSTP and/or RSTP as alternatives to spanning tree?

[Chuck R. Anderson]

On Thu, Apr 28, 2005 at 03:44:47PM -0500, Nicola Foggi wrote:
> We ran across the limited # of spanning tree instances a couple of
> times unfortunately...  we haven't looked at MST yet, but we did end up
> manually pruning the vlan's off the trunked interfaces for vlans that
> were not in use which solved the # of spanning tree.. it also solved a
> problem we were having with the cam table's filling up and caused
> multicast reliability problems...  We looked at enabling Cisco's
> Auto-Pruning functionality, but in our test, because it relies on
> spanning tree to determine if a vlan is needed on a trunk, if you are
> passed the number of supported instances of spanning tree, the 3500 will
> disable spanning tree for a potential vlan that is needed on the
> switch... which then doesn't allow it on the trunk.

Interesting.  Not having used Cisco, I've always manually configured
all the necessary VLANs on the proper ports.  Currently, we have
mostly one Spanning Tree instance everywhere, with by-design loops
that are normally blocked for redundancy.

For redundancy/load-sharing we are moving away from Spanning Tree and
towards Link Aggregation across two core routing switches, utilizing
Nortel's Split Multi-Link Trunking feature.  Then we won't need
802.1d/s/w at all.  We could use 802.3ad, but probably won't, because
static LAG configurations are stateless/protocol-less, and hence more
reliable.

-- 
Chuck R. Anderson			cra at wpi.edu
Network Engineer			(508) 831-6110
Network Operations - Morgan Hall	(508) 831-6666
Worcester Polytechnic Institute	    Fax (508) 831-6669