70 likes | 195 Views
Pseudo-Wire Protection. Ping Pan ppan@hammerheadsystems.com IETF 65. Why?. PW (Important traffic). Tunnel-1 (BW1). Node 2. Node1. PW (Best effort). Tunnel-2 (BW2), backup Tunnel-1. Switch-over. Conventional mechanism (e.g. MPLS FRR) would work, if
E N D
Pseudo-Wire Protection Ping Pan ppan@hammerheadsystems.com IETF 65
Why? PW (Important traffic) Tunnel-1 (BW1) Node 2 Node1 PW (Best effort) Tunnel-2 (BW2), backup Tunnel-1 Switch-over • Conventional mechanism (e.g. MPLS FRR) would work, if • Tunnel-1 and Tunnel-2 have the same bandwidth (i.e.,BW1 = BW2) • All PW’s are equally important • Otherwise, if BW2 < BW1, • Important traffic may get dropped/delayed during network failure • PW protection: • Each PW needs to have its own “importance” and QoS level • During switch-over, preempt others if necessary
Design Considerations (1) • Need to work in single and multi-segment environment • Work with Generic ID FEC only • Working and protection PW’s must be uniquely identified • But make the use of AII/AGI may not be a good idea working protection Need to know what to process
Design Considerations (2) • On Backup Path Determination • Carrying “ERO” should be optional • Use hop-by-hop forwarding for the working PW • Get “RRO” information from “PW Switching Point TLV (SP-TLV)” • Build the protection PW base on SP-TLV • May use SP-TLV for loop-detection • On CAC at each hop • Should be optional depending on traffic type • Other relevant protection information: • 1:1 or 1:N: need to validate the need of 1:N • Fate-sharing: useful in multi-hop applications • Hot-standby vs. cold-standby
Protocol Extension • Protection-TLV • Setup and Holding Preference Levels • Protection Type: hot or cold standby • Schemes: 1:1 or 1:N • Flags: • Fate-sharing allowed • Per-hop BW CAC required • Reference ID: unique identify working and protecting PW’s • PW Status: • Bad message, no bandwidth etc. • Everything else: • Keep as it is • AII/AGI must be the same on all working and protecting PW’s
How does it work? Protecting PW Working PW Node 3 Node 2 Node1 • Run MHOP PW for working and protection PW setup • Use SP-TLV for loop detection • Use Protection-TLV and SP-TLV to forward working and protection PW’s • Forward failure status if not working • Run the Generic FEC procedure to setup the working PW • Run the Generic FEC procedure to setup the protection PW • When working, pointing user traffic to the protecting PW • During failure (learned from BFD etc.), switch over • Preempt traffic when needed… • Send out Label Mapping and setup the working PW • When working, setup the protecting PW (base on SP-TLV…) • When working, pointing user traffic to the protecting PW • During failure (learned from BFD etc.), switch over • Preempt traffic when needed…
Next Step • This proposal works in both single-segment and multi-segment cases • Like to • Continue to gather feedback • Iron out the details with others • Make this into a WG Working Document THANK YOU ALL