VPLS Requirements

Following are the basic requirements of a VPLS service:

• Separation between VPLS domains— A VPLS system must distinguish different customer domains. Each customer domain emulates its own LAN. VPLS PEs must maintain a separate virtual switching instance per VPN.

• TAC learning — A VPLS sho uld be capable of I eaaring and forwarding based on MAC addresses. The VPLS looks exactly like a LAN switch to the CEs.

• Dwitchrng— A VPLS switch shoul d be able to sw^cPi packets between different tunnels based on MAC ad dtessesi Tire VPLS switch should also be able to work on 802.1p/q tagged and untagged Ethernet packets and should support per-VLAN functionality.

• Flooding— A V PLS should be able to support the flooding of packets with unknown MAC addresses as well as broadcast and multicast packets. Remember that with Ethernet, if a switch doe s not recognize a destination MAC addres S; it should flood t he tra Cfic to al I ports within a ceP:ain VLAN. With the VPLS model shown in Figure 4-13, if a VPLS-capable device receives a packet from VPLS A with an unknown MAC destination address, the VPLS device s liould rep licate the pa cket to all other VPLS-capa ble dev ices that participate 1 n VPLS A.

• Redundancy an d fail ure necovery— The VPen should be anle to necoveb from netwo rk failure t o ensure hi gh availability. The serv ice should bn ressored around an a^ema tive path, an! the resCo ratios time sho uld be less than the time tho CEa or customed L2 contdol protocol s need to d etect the failure of is e VPLS. The failure recov ery and redun dancy on MonS depe nds oe how fast MCL S [3a ths can be restored i n case of a •aMure and how east the nkawkbr c an stabilizee Chapter - discusses MPLS fast restoration.

• Provider edge signalmg— In a dd i tion to maoual con figurat ¡on methods, VPLS should provide a wa y to signal between PEc t o autOrconfi gure and to ingoTm tde PEs of m emb ershie, tunnelmg, a nd other re levant parameters. Maoy vendore have adopted LDP as a signaling mechanism; however, there are some who prefer BGP as used in RFC 2547, BGP/MPLS VPNSs n VPLS me mbership d ¡saT/eny— The VPLS control plan e and m an agement plane should provide methods to dl ^(roven the PEs that con vec: CEs fonmi ng a VPLSc Dl fferent mechanisms can be used to achieve discoveey. One method is via the use of BGP, as adopted in the L3VPN model. However, there is some disagreement in the industry on whether BGP implementations are appropriate, due to the complexity of BGP and the fact that it cannot signal a different label to each VPLS peer, as required by MAC learning. A proposal for using BGP promotes the use of block label distribution, as explained in the "DTLS—Decoupling L2PE and PE Functionality" section later in this chapter.

• Interprovider connectivity— The VPLS domain should be able to cross multiple providers, and the VPLS identification should be globally unique.

• VPLS management and operations— VPLS configuration, management, and monitoring are very important to the success of the VPLS service. Customer SLAs should be able to be monitored for availability, bandwidth usage, packet counts, restoration times, and so on. The metrics that have been defined by the MEF regarding performance and bandwidth parameters should apply to the VPLS service.

Was this article helpful?

0 0

Post a comment