[c-nsp] MST spanning-tree

[Steven Fischer]

When we relocated our data center, we opted to deploy MST as the
spanning-tree protocol, given that our data center is almost exclusively
layer 2, we have a lot of vlans, and that number is only going to grow.  We
have two spanning-tree MST instances, 1 and 2, and each contains the vlans
that are either odd (instance 1), or even (instance 2).

It seems that we can create VLANs with no issue, and by default, those VLANs
are placed in the default instance, instance 0.  This really isn't a
problem, but when we move the VLAN into its proper instance, a spanning-tree
recalc appears to occur, the duration of which is long enough to interrupt
data transfers that may be going on at the time.  Other than returning to
PVST/PVST+, is there a way to avoid this recalc taking out half the VLANs in
the data center for 30 seconds when adding it to the proper instance?  Will
our planned migration to VSS mitigate this to any degree, and if so, how
much?

-- 
To him who is able to keep you from falling and to present you before his
glorious presence without fault and with great joy