The MPLS WG Archive

["Connell, Jeff" <JConnell@WhiteRockNetworks.com>]

Lou et all,
	In reading the new GMPLS draft, section 3.1.4 Bandwidth encoding,
you state the PDR and CDR fields be used to specify the maximum reservable
bandwidth.  This needs to be extended to indicate the number of each type
available.  For instance, if I have two links to the same dest., one with 2
STS-12c's available and the other with 1 STS-12c available, I would prefer
the first due to load balancing (all else equal.) Or, what if I wanted to
provision 12 STS-1s, I would only know I have at least one available.  An
advantage of your proposed method is: only when "Signal Type" is no longer
available does the TE TLV need to be re-flooded. If the metrics change with
every provisioned cross-connect, there could be a heavier TE TLV load on the
network. Thresholds can be used to solve that problem.  I think there needs
to be more information than is currently flooded.

	On an editorial note, I notice in this draft (and the earlier
version)  STS-1s are mentioned, and then OC-3, OC-12, etc.. Do you mean
STS-3c, and STS-12c?  I know the intent is to keep the difference between
SONET and WDM (not wavelength, but bandwidth) abstract, but it does make the
SONET guys cringe a bit :) As you know, there does not exist a OC-12 within
an OC-48, but there can be up to four STS-12c's. It would be more correct to
say : STS-12c/OC-12/STM-4 for signal type (SONET/WDM/SDH).

Jeff Connell
White Rock Networks
(972)588-3762
www.whiterocknetworks.com