90 likes | 189 Views
GMPLS Asymmetric Bandwidth Bidirectional LSPs draft-berger-ccamp-asymm-bw-bidir-lsps-00.txt. Lou Berger <lberger@labn.net> Attila Takacs <attila.takacs@ericsson.com> Diego Caviglia <diego.caviglia@ericsson.com> Don Fedyk <dwfedyk@nortel.com> Julien Meuric <julien.meuric@orange-ftgroup.com>.
E N D
GMPLS Asymmetric Bandwidth Bidirectional LSPsdraft-berger-ccamp-asymm-bw-bidir-lsps-00.txt Lou Berger <lberger@labn.net> Attila Takacs <attila.takacs@ericsson.com> Diego Caviglia <diego.caviglia@ericsson.com> Don Fedyk <dwfedyk@nortel.com> Julien Meuric <julien.meuric@orange-ftgroup.com> CCAMP - 69th IETF
Background • Asymmetric LSP Bandwidth is actually an old discussion (role reversal time!) • From: Adrian Farrel <AF@dataconnection.com>To: mpls@UU.NETSubject: Asymmetrical Bi-directional LSPsDate: Wed, 27 Sep 2000 22:33:57 +0100 • Conclusion at that time: no real need for such LSPs • Trigger for draft • Raised in PBB-TE context • Discussed in draft-takacs-asym-bw-lsp-00.txt CCAMP - 69th IETF
Draft Objective • As this conversation keeps resurfacing, we propose that the WG determine and document: • The preferred way to provide this function • Should this be a standard or experimental function CCAMP - 69th IETF
Proposed Solutions • Draft covers two approaches: • Both versions use a single bidirectional LSP • Technology specific • Uses ADSPEC to carry traffic parameters for upstream data • Requires MEF Ethernet Traffic Parameters • Generic • Defines new upstream traffic parameter related object classes to match old downstream objects Propose that WG select and pursue one approach CCAMP - 69th IETF
Some Context • RFC2205 (RSVP) and RFC3209 (RSVP-TE) • Data flow –Ingress Egress • SENDER_TSPEC –Ingress desired transmission characteristics • FLOWSPEC –Egress desired receive/downstream QoS • ADSPEC –What is being provided by network* • RFC3473 (GMPLS-RSVP-TE) • Data flow –Ingress Egress • SENDER_TSPEC –Desired QoS for both upstream and downstream • FLOWSPEC –Confirmation of allocated resources • ADSPEC –Unmodified in general – note none for upstream, unused for some specific technologies |---| Path |---| | I |----------------->| E | | n | -SENDER_TSPEC | g | | g | -ADSPEC | r | | r | | e | | e | Resv | s | | s |<-----------------| s | | s | -FLOWSPEC | | |---| |---| Data Flow -----------------------> |---| Path |---| | I |----------------->| E | | n | -SENDER_TSPEC | g | | g | -ADSPEC | r | | r | | e | | e | Resv | s | | s |<-----------------| s | | s | -FLOWSPEC | | |---| |---| Data Flow <----------------------> CCAMP - 69th IETF
Option 1: Technology Specific |---| Path |---| | I |----------------->| E | | n | -SENDER_TSPEC | g | | g | -ADSPEC | r | | r | | e | | e | Resv | s | | s |<-----------------| s | | s | -FLOWSPEC | | |---| |---| • Per technology solution • Draft defines Ethernet specific solution • ADSPEC carries upstream traffic parameters • Draft requires MEF Ethernet Traffic Parameters • Solution only compatible with technologies that don't require/use ADSPEC • SENDER_TSPEC and FLOWSPEC per RFC2205/3209 CCAMP - 69th IETF
Option 2: Generic |---| Path |---| | I |------------------->| E | | n | -SENDER_TSPEC | g | | g | -ADSPEC | r | | r | -UPSTREAM_FLOWSPEC | e | | e | | s | | s | Resv | s | | s |<-------------------| | | | -FLOWSPEC | | | | -UPSTREAM_TSPEC | | |---| -UPSTREAM_ADSPEC |---| • New upstream objects to match RFC2205 downstream objects • UPSTREAM_FLOWSPEC –Ingress desired receive/upstream QoS • UPSTREAM_TSPEC –Egress desired transmission characteristics • UPSTREAM_ADSPEC –What is being provided by network* • Draft missed handling of UPSTREAM_TSPEC / UPSTREAM_FLOWSPEC mismatches • Next rev will state to use MBB CCAMP - 69th IETF
Opt. 1: Tech. Specific Pros: Simple to implement, i.e., no message format changes Cons: Need change per technology Precludes original use of ADPSEC Only works with technologies that don’t require ADSPEC Opt. 2: Generic Pros: Works with any technology Fixes omission of Upstream ADSPEC introduced in RFC3473 Cons: Three new object classes Alternatives Summary CCAMP - 69th IETF
Next Steps • Answer major questions: • Technical direction: • Per technology solutions (Ethernet today) • Generalized solution • Process direction: • If required functionality Standard • If not Experimental • General draft update • Other comments? CCAMP - 69th IETF