GVPLS/LPE - Generic VPLS Solution based on LPE Framework Update version 01 Vasile Radoaca/Dinesh Mohan Nortel Networks Vasile Radoaca Andrew Malis Ananth Nagarajan Alain Vedrenne Javier Achirica Himanshu Shah Muneyoshi Sujuki Pascal Menezes Yael Dayan Marty Borden Yaron Raz Simon Hunt GVPLS/LPE - 1
Agenda • Characteristics of the GVPLS Model • Key elements of the version 01 • Requirements compliance • Next Steps GVPLS/LPE - 2
History • LPE Framework document – September 2001 • GVPLS/LPE solution - November 2002 • draft-rosen-ppvpn-l2-signaling-02.txt – March 2002 GVPLS/LPE - 3
Characteristics of the GVPLS model • Decomposes MAC Learning and Forwarding from the VPLS core forwarding – MAC learning is done on the U-PE access devices – VPLS core forwarding is done on the N-PE devices • Minimizes the number of control sessions between the U-PE devices – regardless of the number of U-PE devices, each U-PE device has maximum 2 control sessions with the local N-PE device • The VPLS core PWs identify both the N-PE and U-PE devices – tradeoff between MAC-Learning and the number of PWs in the core • Allows efficient OAM and Multicast schemes to be implemented – provides a consistent traceability path between SP U-PE devices GVPLS/LPE - 4
GVPLS Reference Model Attachment VC Extension PWs: VC-6 VC-1 VC-2 VC-7 U-PE U-PE CE PW: P2P or MP2P CE VC-4 N-PE A N-PE B VSI-U VSI-U SET VSI-N VSI-N CE CE VSI VSI VC-8 Attachment VC VC-3 GVPLS/LPE - 5
Key elements for version 01 • Extends and complements the non-distributed model – non-distributed and distributed models can co-exist together • Aligns with the Functional decomposition proposed by the DT • Explains the rationale of the Distributed Model and deployment scenarios • Supports clear separation between access and metro-core networks – forwarding the customer packets is based on a SP Address scheme instead of the customer MAC addresses • Supports different access technologies – Q-in-Q – PW Martini – 802.1 Q VLAN – flexible to adopt new L2 protocols • Supports P2P PW and M2P PW in a consistent way – The M2P PW are not mandatory, but may enhance the scalability of the model • Extends the PWE3/Martini signaling method to support both the distributed and non-distributed models– also, it allows P2P, M2P PWs to coexist together • Allows inter-working between VPLS/HVPLS and GVPLS • Supports L2 VPN Requirements GVPLS/LPE - 6
GVPLS Solution Abstraction Layer Value - Common Information Model - Common Signaling Model - Common Date Plan - Common Auto-discovery model DPE HPE -U-PE-s devices -U-PE-sf devices -Access Encapsulation -Access Encapsulation -Martini -Martini -Q-in-Q -MAC-in-MAC -VLAN -proprietary type -- MPLS core -- MPLS core Solutions Solutions: LPE, DTLS VPLS/HVPLS GVPLS/LPE - 7
Additional Requirements addressed by version 01 • Performance & Scalability transparency to the type & number of CE devices • Scalable deployment: – Incremental build-up – # of VPNs – # of End-points/VPLS – # of customer MAC addresses • VPLS Customers protection – remove Single Point of Failure – isolates the customer MAC addresses to the SP Edge U-PE and forward the customer packets using SP address scheme • Allow the N-PE devices to perform simultaneous multiple services [IPVPN, VPLS, L2VPN] GVPLS/LPE - 8
Future Steps • Convergence on Signaling – draft-rosen-ppvn-signaling-02.txt • Integrate QoS, Resiliency • Enhance OAM capabilities • Convergence with VPLS GVPLS/LPE - 9
History of the Distributed model • LPE Framework document – publish on September 2001 • GVPLS/LPE solution - publish on November 2002 • DTLS – publish August 2001 • draft-rosen-ppvpn-l2-signaling-02.txt – publish on 2002 GVPLS/LPE - 10
Recommend
More recommend
