90 likes | 240 Views
IETF-53 PWE3 Workgroup 21-Mar-02 An Update on CESoPSN draft-vainshtein-cesopsn-02.txt. Sasha Vainshtein, Israel Sasson, Akiva Sadovski, Eduard Metz, Tim Frost. CESoPSN – A Reminder. A PW encapsulation for E3 emulation of TDM circuits The -01 revision first presented at the SLC meeting
E N D
IETF-53PWE3 Workgroup21-Mar-02An Update on CESoPSN draft-vainshtein-cesopsn-02.txt Sasha Vainshtein, Israel Sasson, Akiva Sadovski, Eduard Metz, Tim Frost
CESoPSN – A Reminder • A PW encapsulation for E3 emulation of TDM circuits • The -01 revision first presented at the SLC meeting • Supports: • N*DS0 circuits (with or without CE application signaling) • Unstructured E1 and T1 circuits • Unstructured E3 and DS3 circuits • Complies with the Minimal Intervention principle regarding the TDM payload • Uses RTP for: • Timing (synchronization) services • Sequencing services • Differentiation between data and signaling • PSN-invariant Stray Packets detection • Optionally may use a (payload) control word • Carries PE state in both directions
-02 version • Posted before the Minneapolis IETF after a substantial re-write • A special “Thank you” to Stephen Casner for reviewing the draft before posting: • Corrected some serious errors • Provided important inputs for the future releases • Thanks for discussion of the pre-posting version and valuable feedbacks to: • Sim Narasimha • Max Riegel • Yaron Raz
Major Changes from –01 • An explicit “Requirements” section has been added: • Refers to applicable generic PWE3 requirements (draft-ietf-pwe3-requirements-02.txt) • Defines specific requirements for edge-to-edge emulation of TDM circuits • Support for CE application state signaling has been defined • Channel-Associated Signaling (CAS) is a typical example • “Data packets” and “signaling packets” are defined • Data packets format is not affected by the signaling scheme • Signaling packets may use the “slow path” • Wrt to RTP, the signaling packets use their own PT, SSRS ID and sequencing • Egress resynchronizes data and signaling using RTP timestamps • Only application state transitions are signaled • Payload format is signaling scheme-specific • For CAS, the application state encoding is based on RFC 2833 • Protection against loss of signaling packets and state recovery based on RFC 2833
More Changes from -01 • PW (VC) Types and and RTP Payload Types have been decoupled • Dynamic RTP Payload Types used instead • G.826-compatible Performance Monitoring has been defined • Minor: • Text aligned with the latest revisions of the common PWE3 documents • Some non-specific stuff removed • Some descriptive text added on: • Detection of lost packets • Adaptation of the jitter buffer size etc. • In-band loopback commands have been removed • Can be set/released manually or signaled via a common PWE3 maintenance protocol • Loopback PE state still signaled back across the PSN
Open Issues • Improve encapsulation efficiency with minimal loss of functionality. Options: • Skip the control word without loosing functionality? • Optionally use only the control word (use the reserved bits)? • Define a more effective encapsulation for T1? • RTCP usage. Natural possibilities: • Resynchronization of independent SSRCs for data and signaling packets • Retrieval of “far-end” PE state, statistics etc. • Support Common Channel Signaling (CCS) schemes: • Same basic principles as for CAS • Resolve specific encoding issues
Convergence Issues • CESoPSN vs. TDMoIP (draft-anavi) • Almost 100% overlap in functionality • Two meetings have been hold in Tel-Aviv • We seemed to be close, but did not succeed to resolve the differences • CESoPSN vs. CEP (draft-pate) • Some overlap in functionality • Some applications covered only by one of the two documents • No serious attempt to resolve the differences yet • In both cases differences in the payload formats are more difficult to overcome than differences in the header formats • The input from the WG would be helpful • Virtually nothing seen on the list
Questions to the WG and ADs • Requirements for edge-to-edge emulation of TDM circuits: • Is an explicit specification needed? • People use to say that “TDM is something else” both on the list and in private discussions • Being more specific about what this “something else”is could be helpful • If needed, how it should be done: • Within the common PWE3 requirements document? • In a dedicated document? • Within the document that describes a specific solution? • RTP in PWE3: • Is “classic” RTP/UDP/IP admissible in the PWE3 context? • Suits existing infrastructure (e.g., header compression) • How should the definitions of RTP/L2TPv3/IP and RTP/MPLS stacks be pursued? • Can CESoPSN be adopted as a WG Item?