AW: LCAS and draft-mannie-ccamp-gmpls-lbm-tdm-00.txt

["Roese, Josef" <[e-mail]>]

Eric, Dimitri

It took same time to read all the emails about LMB versus LCAS. I will not restart the discussion on pro and con of LMB and LCAS. In all the emails I missed a network and operator view.

For the following scenario I assume that LMB works hitlessly.

GMPLS, ASON, LMB and LCAS will be introduced in the network. Operators make their choice which technology they want to use inside their domain. Later, circuit switched services will be also used between operators. Interworking between domains using different signalling and routing protocols can be achieved by translation to inter-domain protocols (e. g. BGP). This is a proven technique. With such techniques we can easily move from local switched services to global ones, when we exclude virtual concatenated circuits. For sure, some ends will use LCAS others LMB. Sinks that operate both LCAS and LMB protocol may achieve interworking between circuit ends. However, first generation circuit ends will use either LCAS or LMB and will, therfore, not inter-work. A translation of a real time LCAS to a real time LMB protocol and vice versa at domain borders will not work, too.

Conclusion: Different end to end protocols for bandwidth adjustment is the worst solution we could get in a network. Whether LMB may be simpler or more powerful than LCAS is not the issue. In any case we would loose with the introduction of a second protocol that is incompatible with LCAS.


Josef Roese
T-Systems Nova
Technologiezentrum, E531g
Am Kavalleriesand 3, 6429 5 Darmstadt
Tel: [phone]
Fax: [phone]
mailto:[e-mail]
http://www.t-nova.de