VPLS

4,374 views

Published on

Cloud computing environments need to be extended across multiple data center locations to facilitate efficient resource utilization.

VPLS is the prime technology to achieve flexible L2 connectivity between many sites.

This session looks at the details of VPLS.

Published in: Technology
0 Comments
10 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
4,374
On SlideShare
0
From Embeds
0
Number of Embeds
3
Actions
Shares
0
Downloads
419
Comments
0
Likes
10
Embeds 0
No embeds

No notes for slide
  • Private WAN versus Internet WAN for branch Internet for Remote Access DC InterconnectEdge Services TierProvides all WAN services at the edge of data center networks; connects to the WANservices in other locations, including other data centers, campuses, headquarters, branches, carrierservice providers, managed service providers and even cloud service providers.
  • L2 connectivity for VMotion, Cloud TransparencyNo tight L3 boundaries possible
  • Only configure the access ports – not the “switch fabric”
  • Virtualization
  • Only configure the access ports – not the “switch fabric”
  • VPLS

    1. 1. Vpls<br />Bruno De Troch<br />Tech Lead Carrier Ethernet Solutions EMEA<br />October 2010<br />
    2. 2. Agenda<br />Cloud Interconnect Requirements<br />MEF Ethernet Services<br />MPLS Based Services Framework<br />MPLS Basics<br />VPLS and Standardization<br />VPLS Details<br />VPLS and Site Redundancy<br />Summary<br />
    3. 3. Interconnections in the cloud<br />
    4. 4. cloud computing infrastructure<br />Interconnect at scale<br />Within the DC<br />Between DCs in the same metro<br />Between DCs in different metros/countries<br />For your own infrastructure<br />Or as a service to others<br />Desire to have a single technology at most scales<br />Intra-DC and inter-DC<br />Reduction of complexity: minimal or no configuration<br />Re-use of existing infrastructure, operational methodology, systems<br />VLANs and MAC flexibility and scaling<br />Virtualization of infrastructure<br />
    5. 5. DC Edge connectivity – wan access<br />Edge<br />WAN<br />ISP<br />BACKUP DC<br />Branch<br />PRIMARY DC<br />Services<br />Edge<br />IPSec<br />Acceleration<br />Branch<br />Secure access<br />SSL<br />Core/Aggregation<br />Mobile user<br />
    6. 6. Edge connectivity requirements – L2<br />Edge<br />VLAN<br />VLAN<br />WAN<br />ISP<br />BACKUP DC<br />Branch<br />PRIMARY DC<br />Services<br />Edge<br />IPSec<br />Acceleration<br />Branch<br />Secure access<br />SSL<br />Aggregation<br />Mobile user<br />
    7. 7. Model wan as a switch<br />WAN<br />VLAN<br />VLAN<br />VLAN<br />VLAN<br />VLAN<br />L2 SWITCH<br />Edge<br />Edge<br />Edge<br />Edge<br />
    8. 8. Edge connectivity requirements – Also L3<br />Edge<br />VLAN<br />VLAN<br />WAN<br />ISP<br />BACKUP DC<br />Branch<br />PRIMARY DC<br />Services<br />Edge<br />IPSec<br />Acceleration<br />Branch<br />Secure access<br />SSL<br />IP<br />IP<br />IP<br />IP<br />IP<br />Aggregation<br />Mobile user<br />
    9. 9. Model wan as a switch/ROUTER<br />WAN<br />VLAN1<br />VLAN1<br />IP1<br />VLAN1<br />IP1<br />VLAN2<br />SWITCH/ROUTER<br />VLAN1<br />IP2<br />Edge<br />Edge<br />Edge<br />Edge<br />VLAN2<br />(Virtualized WAN)<br />VLAN1<br />IP2<br />IP1<br />IP2<br />VLAN1<br />IP1<br />VLAN2<br />VLAN1<br />VLAN1<br />IP2<br />
    10. 10. Mefethernet services (l2)<br />
    11. 11. Ethernet Services<br />Subscriber Site <br />Subscriber Site <br />MEF describes Ethernet Services<br />E-Line - Point-to-Point<br />E-LAN - Multipoint-to-Multipoint<br />E-Tree - Point-to-Multipoint<br />Ethernet Services “Eth” Layer<br />Service Provider 1<br />Metro Ethernet Network<br />Service Provider 2<br />Metro Ethernet Network<br />Subscriber Site <br />Subscriber Site <br />ETH<br />UNI-N<br />ETH<br />UNI-N<br />ETH<br />UNI-C<br />ETH<br />UNI-N<br />ETH<br />UNI-N<br />ETH<br />UNI-C<br />ETH<br />UNI-N<br />ETH<br />UNI-N<br />
    12. 12. Mpls based Services framework<br />
    13. 13. Leveraging Common Technologiesfor Multiple Applications and Services<br />Ethernet and IP as services<br />Optimized bandwidth scaling<br />Ubiquitous and familiar technology<br />MPLS as a transport technology<br />Resilient<br />Scalable: for both unicast and multicast<br />Truly multi-service<br />Consistent with core network<br />IP control plane<br />Ease of provisioning<br />Scalable<br />Common to all the services<br />Results In Operational Simplicity and Enables Network Convergence<br />
    14. 14. mpls Converged Services Framework<br />E-LAN<br />E-LINE<br />Multicast<br />Ethernet<br />Legacy<br />IP<br />Any L1/L2<br />VPLS<br />TDM<br />FR/ATM<br />L2VPN<br />MPLS<br />IP<br />IPVPN<br />E-Line built using L2VPN<br />E-LAN built using VPLS (Virtual Private LAN Switching)<br />E-Tree built using P2MP or VPLS<br />
    15. 15. Common control-plane<br />E-LAN<br />E-LINE<br />MPLS Multicast<br />Ethernet<br />Legacy<br />IP<br />VPLS<br />TDM<br />FR/ATM<br />L2VPN<br />MPLS<br />IP<br />IPVPN<br />Same operational procedures<br />Legacy, Ethernet & IP<br />Re-use existing skills: trained personnel advantage<br />BGP auto-discovery<br />BGP signaling<br />LDP, RSVP or both<br />
    16. 16. Mpls basics<br />
    17. 17. MPLS is over 10 Years Old<br />Early attempts from Tag Switching to MPLS<br />Tag Switching<br />Date of Birth - March 1996 (during IETF 35); Los Angeles, California, USA<br />Tag Switching Architecture - Sep 1996<br />Tag Distribution Protocol (TDP) – Sep 1996<br />Use of Tag Switching With ATM – Oct 1996<br />Tag Stack Encoding – Nov 1996<br />Multicast Tag Binding and Distribution using PIM – Dec 1996<br />Carrying Tags in RSVP – Dec 1996<br />December 1996 (IETF 37) – MPLS BOF<br />April 1997 (IETF 38) – first MPLS WG meeting<br />Agreeing on the name of the WG was one of the major challenges<br />
    18. 18. MPLS Terminology<br />Label<br />Short, fixed-length packet identifier<br />Unstructured<br />Link local significance<br />Forwarding Equivalence Class (FEC)<br />Stream/flow of IP/Any packets:<br />Forwarded over the same path<br />Treated in the same manner<br />Mapped to the same label<br />FEC/label binding mechanism<br />Currently based on destination IP address prefix<br />Future mappings based on SP-defined policy<br />
    19. 19. MPLS Terminology<br />25<br />IP<br />(1, 22)<br />(2, 17)<br />(1, 24)<br />(3, 17)<br />(1, 25)<br />(4, 19)<br />19<br />IP<br />(2, 23)<br />(3, 12)<br />Connection Table<br />In<br />(port, label)<br />Out<br />(port, label)<br />Label<br />Operation<br />Port 1<br />Port 2<br />Swap<br />Swap<br />Swap<br />Port 3<br />Port 4<br />Swap<br />Label Swapping<br />Connection table maintains mappings<br />Exact match lookup <br />Input (port, label) determines:<br />Label operation<br />Output (port, label)<br />Same forwarding algorithm used in Frame Relay and ATM<br />
    20. 20. MPLS Terminology<br />New York<br />San<br />Francisco<br />LSP<br />Label-Switched Path (LSP)<br />Simplex L2 tunnel across a network<br />Concatenation of one or more label switched hops<br />Analogous to an ATM or Frame Relay PVC<br />
    21. 21. MPLS Terminology<br />LSR<br />New York<br />LSR<br />LSR<br />San<br />Francisco<br />LSR<br />LSP<br />Label-Switching Router (LSR)<br />Forwards MPLS packets using label-switching<br />Capable of forwarding native IP packets<br />Executes one or more IP routing protocols<br />Participates in MPLS control protocols <br />
    22. 22. MPLS Terminology<br />Egress<br />LSR<br />Ingress<br />LSR<br />New York<br />Transit<br />LSR<br />San<br />Francisco<br />Transit<br />LSR<br />LSP<br />Ingress LSR (“head-end LSR”)<br />Examines inbound IP packets and assigns them to an FEC<br />Generates MPLS header and assigns initial label<br />Transit LSR<br />Forwards MPLS packets using label swapping<br />Egress LSR (“tail-end LSR”)<br />Removes the MPLS header<br />
    23. 23. MPLS Header<br />TTL<br />Label (20-bits)<br />CoS<br />S<br />IP Packet<br />L2 Header<br />MPLS Header<br />32-bits<br />Fields<br />Label<br />Experimental (CoS)<br />Stacking bit<br />Time to live<br />IP packet is encapsulated by ingress LSR<br />IP packet is de-encapsulated by egress LSR<br />
    24. 24. MPLS Forwarding Model<br />Source<br />Egress<br />LSR<br />Paris<br />Ingress<br />LSR<br />Rome<br />Ingress LSR determines FEC and assigns a label<br />Forwards Paris traffic on the Green LSP<br />Forwards Rome traffic on the Blue LSP<br />Traffic is label swapped at each transit LSR<br />Egress LSR <br />Removes MPLS header <br />Forwards packet based on destination address<br />
    25. 25. MPLS Forwarding vs. IP Routing<br />Source<br />Destination<br />IP Routing Domain<br />Examine IP header<br />Assign to FEC<br />Forward<br />Examine IP header<br />Assign to FEC<br />Forward<br />Examine IP header<br />Assign to FEC<br />Forward<br />Examine IP header<br />Assign to FEC<br />Forward<br />Egress<br />LSR<br />Ingress<br />LSR<br />Source<br />Destination<br />MPLS Domain<br />Examine IP header<br />Assign to FEC<br />Forward<br />Examine IP header<br />Assign to FEC<br />Forward<br />Label swap<br />Forward<br />Label swap<br />Forward<br />
    26. 26. MPLS Forwarding Example <br />MPLS Table<br />134.5.6.1<br />In<br />Out<br />(2, 84)<br />(6, 0)<br />134.5.1.5<br />99<br />200.3.2.7<br />0<br />200.3.2.7<br />56<br />200.3.2.7<br />2<br />6<br />Egress Routing Table<br />Destination<br />Next Hop<br />2<br />200.3.2.7<br />200.3.2.7<br />134.5/16<br />134.5.6.1<br />3<br />200.3.2/24<br />200.3.2.1<br />1<br />2<br />3<br />5<br />Ingress Routing Table<br />Destination<br />Next Hop<br />134.5/16<br />(2, 84)<br />200.3.2/24<br />(3, 99)<br />MPLS Table<br />MPLS Table<br />200.3.2.7<br />200.3.2.1<br />In<br />Out<br />In<br />Out<br />(1, 99)<br />(2, 56)<br />(3, 56)<br />(5, 0)<br />
    27. 27. How Is an LSP Established?<br />Requires a signaling protocol to:<br />Coordinate label distribution<br />Explicitly route the LSP<br />Bandwidth reservation (optional)<br />Class of Service (DiffServ style)<br />Resource re-assignment<br />Pre-emption of existing LSPs<br />Loop prevention<br />MPLS signaling protocols<br />Label Distribution Protocol (LDP)<br />Resource Reservation Protocol (RSVP)<br />Constrained Routing with LDP (CR-LDP)<br />
    28. 28. MPLS Signaling Protocols<br />The IETF MPLS architecture does not assumea single label distribution protocol<br />LDP<br />Executes hop-by-hop<br />Selects same physical path as IGP<br />Does not support traffic engineering<br />RSVP<br />Easily extensible for explicit routes and label distribution<br />Allows very fast convergence<br />
    29. 29. VPLS and standardization<br />
    30. 30. RFC 4761<br />(BGP)<br />RFC 4762 <br />(LDP)<br />BGP RR<br />T-LDP<br />New node<br />New service<br />Existing control-plane session<br />New control-plane session<br />New node<br />New service<br />Virtual Private LAN ServicesTwo Deployed Standards<br /> LDP-based<br /><ul><li>Signaling only, no auto-discovery
    31. 31. High-touch provisioning</li></ul> BGP-based<br /><ul><li>Signaling & Auto-discovery
    32. 32. Inter-area/ metro/provider
    33. 33. Multicast optimization</li></li></ul><li>VPLS details<br />
    34. 34. VPN A<br />Site2 <br />VPN A<br />Site 1<br />CE–A2<br />VPN B<br />Site2<br />CE–A1<br />P<br />P<br />PE 2 <br />PE 1<br />CE–B2<br />VPN B<br />Site 1<br />VPN A<br />Site 3<br />P<br />P<br />PE 3 <br />CE–B1<br />CE–A3<br />Virtual Private LAN Service<br />Make MPLS network look like an Ethernet switch/hub/wire segment to the edge devices<br />Depends on how much is emulated<br />Edge devices do not see the core network providing VPLS<br />
    35. 35. How Is VPLS Different from P2P?<br />In “point-to-point” Layer 2 VPNs, the layer 2 address identifies a path<br />In VPLS, the layer 2 address identifies a destination endpoint<br />The path to such this destination (i.e., the label stack to be used to get there) must be learned<br />This process is very much like what learning bridges do!<br />I.e., static mapping vs. dynamically learned<br />
    36. 36. Point-to-point Ethernet Access<br />CE 2<br />PE 2<br />VLAN<br />122<br />LSPs<br />PE 1<br />CE 1<br />VLAN<br />122<br />CE 3<br />VLAN<br />123<br />VLAN<br />123<br />PE 3<br />Customer frames are switched based on VLAN tag<br />Each VLAN from a CE identifies a remote CE<br />The (outer+inner) label stack is essentially a continuation of the Ethernet VLAN<br />If a frame sent on VLAN 122 goes to CE x, then a frame received on VLAN 122 comes from CE x<br />Customer appears to have two independent VLANs<br />
    37. 37. Multipoint Ethernet Access<br />CE 2<br />PE 2<br />LSPs<br />PE 1<br />CE 1<br />CE 3<br />PE 3<br />Customer frames are switched on MAC addresses<br />Single (or no) VLAN from CE to PE<br />PE must choose the (outer+inner) label stack based on the destination MAC address<br />Customer appears to have a single broadcast domain (LAN) connecting all the sites<br />
    38. 38. VPLS Operation<br />Sending to an unknown MAC address<br />“Flood” to all members of the VPLS<br />Sending to a known MAC address<br />Mapping to <outer label, inner label> exists<br />Receiving from some MAC address y<br />Identify the sender; find the label stack that will reach that sender, and map MAC address y to that label stack in the MAC address cache<br />Periodically, age out unused entries from the MAC address cache<br />
    39. 39. VPLS Building Blocks<br />Site Discovery<br />Signaling<br />MAC Forwarding<br />Flooding<br />Learning<br />Aging<br />
    40. 40. VFT/MAC Cache for a VPLS<br />y<br />flood<br />VPN A<br />Site2 <br />x<br />VPN A<br />Site 1<br />CE–A2<br />VPN B<br />Site2<br />CE–A1<br />P<br />P<br />PE 2 <br />PE 1<br />CE–B2<br />VPN B<br />Site 1<br />VPN A<br />Site 3<br />P<br />P<br />PE 3 <br />y<br />CE–B1<br />CE–A3<br />PE1’s VFT for VPN A<br />MAC address cache<br />
    41. 41. VFT/MAC Cache for a VPLS<br />VPN A<br />Site2 <br />x<br />VPN A<br />Site 1<br />CE–A2<br />VPN B<br />Site2<br />CE–A1<br />P<br />P<br />PE 2 <br />PE 1<br />CE–B2<br />VPN B<br />Site 1<br />VPN A<br />Site 3<br />P<br />P<br />PE 3 <br />CE–B1<br />y<br />CE–A3<br />PE1’s VFT for VPN A<br />MAC address cache<br /> y 654 3001<br />Pkt arrives with src MAC addr y and recv label 1003<br />
    42. 42. Why Full Mesh the PEs?<br />CE 2’<br />PE 2<br />LSPs<br />PE 1<br />CE 1<br />CE 2<br />CE 3<br />PE 3<br />Flood packet from a CE to all other PEs<br />Unknowns, broadcasts, multicasts<br />Flood packet from core to all of PE’s CEs<br />Never send (flood or forward) packet from the core back to the core! Thus, no loops among PEs<br />
    43. 43. Site 2<br />Site 3<br />PE-2<br />PE-1<br />CE-3<br />CE-2<br />Vlan 10<br />Vlan 10<br />VFT<br />VFT<br />Site 3 VCT<br />Site 2 VCT<br />100:1.2.3.3<br />100:1.2.3.2<br />Route Dist<br />Route Dist<br /> LSP<br /> LSP<br />640<br />320<br />3<br />2<br />VE ID<br />VE ID<br />20<br />20<br />Sites<br />Sites<br />Label base<br />Label base<br />Route Target<br />RED<br />Route Target<br />RED<br />PE VCT Provisioning<br />3000<br />2000<br />For VPLS Domain RED<br />PE-1 is configured with Site 2 VCT<br />PE-2 is configured with Site 3 VCT<br />Each PE automatically allocates a VPN label block to be used as de-multiplexors<br />
    44. 44. Site 2<br />Site 3<br />PE-3<br />PE-2<br />CE-3<br />CE-2<br />Vlan 10<br />Vlan 10<br />VFT<br />VFT<br />PE-2’s VFT for VPLS RED<br /> LSP<br /> LSP<br />640<br />320<br />Outer<br />Inner RX<br />VE-ID<br />Inner TX<br />1<br />2001<br />100:1.2.3.2<br />Route Dist<br />3<br />VE ID<br />2<br />Label used by site 3 to reach Site 2<br />.<br />.<br />.<br />.<br />20<br />Sites<br />Used by PE-2 to do MAC learning from site 3<br />20<br />2020<br />2000<br />Label base<br />Route Target<br />RED<br />2003<br />VPLS Forwarding Table<br />Site 2 VCT<br />VPN Forwarding Table (VFT) on PE holds all the VCTs information<br />Also contains MAC forwarding information (FDB)<br />
    45. 45. VPLS Auto-discovery & Signaling<br />MP-iBGP<br />Site 2<br />Site 3<br />PE-3<br />PE-2<br />CE-3<br />CE-2<br />Vlan 10<br />Vlan 10<br />VFT<br />VFT<br />Site 3 VCT NLRI<br />Site 2 VCT NLRI<br />100:1.2.3.3<br />Route Dist<br />100:1.2.3.2<br />Route Dist<br /> LSP<br /> LSP<br />3<br />2<br />VE ID<br />VE ID<br />640<br />320<br />Sites<br />20<br />20<br />Sites<br />3000<br />Label base<br />2000<br />Label base<br />RED<br />Route Target<br />Route Target<br />RED<br />PE-2<br />Next Hop<br />PE-3<br />Next Hop<br /><ul><li>PE-PE VCT distribution using Multi-Protocol BGP (RFC 2858)
    46. 46. Requires full-mesh MP-iBGP or Route Reflectors
    47. 47. Route Distinguisher: “uniquifies” VCT information
    48. 48. Route Target: determines VPN topology
    49. 49. Analogous to CE-PE routes advertisements in RFC2547 VPNs
    50. 50. One single LNRI advertisement per VPLS instance per PE is sufficient</li></li></ul><li>VPLS Auto-discovery & Signaling<br />MP-iBGP<br />Site 2<br />Site 3<br />PE-3<br />PE-2<br />CE-3<br />CE-2<br />Vlan 10<br />Vlan 10<br />VFT<br />VFT<br />Site 2 VCT NLRI<br />Site 3 VCT NLRI<br />100:1.2.3.2<br />100:1.2.3.3<br />Route Dist<br />Route Dist<br /> LSP<br /> LSP<br />2<br />3<br />VE ID<br />VE ID<br />640<br />320<br />20<br />20<br />Sites<br />Sites<br />3000<br />Label base<br />Label base<br />2000<br />Route Target<br />RED<br />Route Target<br />RED<br />PE-2<br />Next Hop<br />PE-3<br />Next Hop<br />Label used to reach site 3<br />640<br />3002<br />PE-2’s VFT for VPLS RED<br />outer<br />Inner RX<br />VE-ID<br />Inner TX<br />1<br />3<br />2003<br />.<br />.<br />.<br />.<br />20<br />PE-2 receives BGP NLRI from PE-3’s for RED VPLS instance site 3<br />
    51. 51. VPLS Auto-discovery & Signaling<br />MP-iBGP<br />Site 2<br />Site 3<br />PE-3<br />PE-2<br />CE-3<br />CE-2<br />Vlan 10<br />Vlan 10<br />VFT<br />VFT<br /> LSP<br /> LSP<br />640<br />320<br />PE-2’s VFT for VPLS RED<br />PE-3’s VFT for VPLS RED<br />outer<br />Inner RX<br />outer<br />Inner RX<br />VE-ID<br />VE-ID<br />Inner TX<br />Inner TX<br />1<br />1<br />2001<br />3001<br />600<br />300<br />5002<br />5003<br />3<br />2<br />640<br />3002<br />2003<br />320<br />2003<br />3002<br />.<br />.<br />.<br />.<br />.<br />.<br />.<br />.<br />15<br />15<br />2020<br />9002<br />3020<br />9003<br />670<br />360<br /><ul><li>A full mesh of pseudo-wires are set-up between all VPLS instances for VPLS RED</li></li></ul><li>Site 2<br />Site 3<br />PE-3<br />PE-2<br />CE-3<br />CE-2<br />X<br />Vlan 10<br />Vlan 10<br />VFT<br />VFT<br />VC label2003<br /> LSP<br /> LSP<br />640<br />320<br />Outer<br />Inner RX<br />VE-ID<br />Inner TX<br />1<br />3001<br />300<br />5003<br />2<br />3002<br />320<br />2003<br />.<br />.<br />.<br />.<br />20<br />3020<br />9003<br />360<br />VPLS MAC Learning:Forwarding to an Unknown MAC Address<br />X sends a packet<br />Tunnel label 320<br />VC label2003<br />L2 Ethernet Frame with Source MAC X<br />Minus preamble, minus checksum<br />L2 Ethernet Frame with Source MAC X<br />Minus preamble, minus checksum<br />PE-3’s VFT for VPLS RED<br />If the destination address is unknown, the packet is “Flooded” to the VPLS domain<br />‘Split Horizon’ forwarding scheme<br />
    52. 52. Site 2<br />Site 3<br />PE-3<br />PE-2<br />CE-3<br />CE-2<br />X<br />Vlan 10<br />Vlan 10<br />VFT<br />VFT<br />VC label2003<br />Tunnel label 320<br /> LSP<br /> LSP<br />640<br />320<br />VC label2003<br />L2 Ethernet Frame with Source MAC X<br />Minus preamble, minus checksum<br />PE-2’s VFT for VPLS RED<br />PE-2’s VPLS RED FDB<br />outer<br />Inner RX<br />outer<br />VE-ID<br />MAC<br />Inner TX<br />Inner TX<br />1<br />X<br />2001<br />600<br />640<br />5002<br />3002<br />3<br />640<br />3002<br />2003<br />.<br />.<br />.<br />.<br />.<br />.<br />.<br />20<br />2020<br />9002<br />670<br />VPLS MAC Learning:Forwarding to an Unknown MAC Address<br />X sends a packet<br />L2 Ethernet Frame with Source MAC X<br />Minus preamble, minus checksum<br />The ‘VC label’ received by PE-2 defines<br />On which VPLS instance the MAC lookup should be done<br />On which site the source MAC address being received resides<br />
    53. 53. Site 2<br />Site 3<br />PE-3<br />PE-2<br />CE-3<br />CE-2<br />X<br />Z<br />Y<br />Vlan 10<br />Vlan 10<br />Unicast to MAC X<br />VFT<br />VFT<br />Tunnel label 640<br /> LSP<br /> LSP<br />VC label3002<br />640<br />320<br />VC label3002<br />L2 Ethernet Frame with Dest MAC X<br />L2 Ethernet Frame with Dest MAC X<br />PE-2’s VFT for VPLS RED<br />PE-2’s VPLS RED FDB<br />outer<br />Inner RX<br />outer<br />VE-ID<br />MAC<br />Inner TX<br />Inner TX<br />1<br />X<br />X<br />2001<br />600<br />640<br />640<br />5002<br />3002<br />3002<br />3<br />Y<br />640<br />3002<br />2003<br />640<br />3002<br />.<br />.<br />.<br />.<br />.<br />.<br />.<br />20<br />P<br />2020<br />9002<br />9002<br />670<br />670<br />VPLS MAC Learning:Forwarding to a Known MAC Address<br /><ul><li>Sending to a known MAC addressX
    54. 54. Two labels derived from FDB lookup
    55. 55. “Martini” Encapsulation</li></li></ul><li>Site 2<br />Site 3<br />PE-3<br />PE-2<br />CE-3<br />CE-2<br />X<br />Z<br />Y<br />Vlan 10<br />Vlan 10<br />VFT<br />VFT<br /> LSP<br /> LSP<br />640<br />320<br />PE-2’s VFT for VPLS RED<br />PE-2’s VPLS RED FDB<br />outer<br />Inner RX<br />outer<br />VE-ID<br />MAC<br />Inner TX<br />Inner TX<br />1<br />X<br />2001<br />600<br />640<br />5002<br />3002<br />3<br />.<br />640<br />3002<br />2003<br />.<br />.<br />.<br />.<br />.<br />.<br />.<br />.<br />.<br />20<br />.<br />2020<br />9002<br />.<br />670<br />.<br />VPLS MAC Aging<br /><ul><li> Periodically age out unused entries from the MAC address cache
    56. 56. MAC address cache should be limited by VPLS instance</li></li></ul><li>VPLS and site redundancy<br />
    57. 57. PE 2<br />Loop scenarios<br />L2<br />PE 1<br />Multi-homedCE<br />L2<br />VPLS<br />L2<br />PE 3<br />PE 2<br />L2<br />PE 1<br />Multi-homedSite<br />L2<br />L2<br />VPLS<br />L2<br />PE 3<br />
    58. 58. PE 2<br />Loop avoidance:spanning tree<br />SPT<br />PE 1<br />Multi-homedCE<br />L2<br />VPLS<br />SPT<br />PE 3<br />PE 2<br />SPT<br />PE 1<br />Multi-homedSite<br />SPT<br />L2<br />VPLS<br />SPT<br />PE 3<br />
    59. 59. Loop avoidance:vpls multi-homing<br />STANDBY<br />PE 2<br />Site-id X<br />BGP<br />Selection*<br />PE 1<br />Multi-homedCE<br />VPLS<br />Site-id X<br />PE 3<br />ACTIVE<br />* BGP Selection based on Local Preference<br />
    60. 60. Model wan as a redundant switch<br />Site-id X<br />Site-id X<br />Site-id Y<br />Site-id Y<br />WAN<br />VLAN<br />VLAN<br />VLAN<br />VLAN<br />VLAN<br />L2 SWITCH<br />Edge<br />Edge<br />Edge<br />Edge<br />Site-id Z<br />
    61. 61. summary<br />
    62. 62. summary<br />VPLS Satisfies the L2 Inter-DC requirements<br />Model the WAN as a switch<br />Allows redundant connections<br />It is coherent with other virtualization techniques<br />MPLS as the unifying layer<br />L3VPN is using same mechanism<br />All mechanisms are standardized<br />

    ×