70 likes | 193 Views
VCAT/LCAS and GMPLS. draft-ietf-ccamp-gmpls-vcat-lcas-01.txt Greg Bernstein (Grotto Networking) Diego Caviglia (Ericsson) Richard Rabbat (Google) Huub van Helvoort (Huawei) IETF 68: Prague. Process. Main editor reverts to Greg Bernstein Large, ad hoc discussion mailing list
E N D
VCAT/LCAS and GMPLS draft-ietf-ccamp-gmpls-vcat-lcas-01.txt Greg Bernstein (Grotto Networking) Diego Caviglia (Ericsson) Richard Rabbat (Google) Huub van Helvoort (Huawei) IETF 68: Prague
Process • Main editor reverts to Greg Bernstein • Large, ad hoc discussion mailing list • Greg Bernstein (Grotto Networking) • Diego Caviglia (Ericsson) • Richard Rabbat (Google) • Huub van Helvoort (Huawei) • Julien Meuric (France Telecom) • Deborah Brungard (at&t) • Lyndon Ong (Ciena) • Jonathan Sadler (Tellabs) • Stephen Shew (Nortel) • Jim Jones (Alcatel-Lucent) • Dan Li (Huawei) • Evelyne Roch (Nortel) • Fred Gruman (Fujitsu) • Vijay Pandian (Sycamore) • Adrian Farrel (Old Dog Consulting) • To be involved in the discussions, comment on the CCAMP list or send us email
Progress and Plans • No update to the draft since last meeting • Plenty of discussion • Reaching agreement on the requirements • Working out what the solution might be • Planning a new revision soon(!) • Completely stabilise the requirements • First draft of solution • Will be ready for review • We should liaise this to ITU-T and OIF who are interested in this work
Requirements • Allow TDM LSPs to be signalled to support virtual concatenation • RFC 4606 defines mechanisms if all VCG components follow the same path (i.e., a single control plane LSP) • Hardware now allows VCG components to follow different paths (i.e., separate control plane LSPs) • Need mechanisms for… • Associating LSPs to form a single group • Defining the characteristics of the group • Size of group (i.e., capacity) • Granularity of group members • Number of group members • A pre-established pool of ‘spare’ LSPs that can be added to a VCG • Determining which ports/interfaces at the egress can support the VCG
Moving Towards a Solution • Association could be though: • Association object • GMPLS Call • Seems like a more natural construct given that the association is only for end points • VCG characteristics can be Call parameters • Management of a spare pool needs some more thought, but there are some ideas • This function needs LCAS to work • Use a Call per “pool” • Use LCAS to manage the VCGs
Determining End-Point Capabilities • Feels like a routing function • But it doesn’t scale • Consider VCG may be spread across ports • Perhaps put some basic features as TE node capabilities in routing • Diverse path VCG support • Perhaps put some basic features as TE link capabilities in routing • LCAS support • Remaining function is in Call “negotiation”