bgp based multi-homing in virtual private lan service wim henderickx florin balus
TRANSCRIPT
Problem Statement
CE1 VPLS
PE1
PE2
PE3
PE4
CE1 is multi-homed to PE1 and PE2; CE1 wants a resilient connectivity
Simple dual connectivity leads to loops and duplicate packets. Proposal provides a complementary solution next to STP, MC-LAG, etc leveraging BGP
CE3
CE4
Goal• Address both PE and access link failure• Provides fast convergence times• Only the traffic transiting the affected network
elements should be impacted• Decouple the Multi-Homing mechanism from the
PW signaling• Minimize the traffic load on the network; ideally just
the local PEs should be involved in the selection process
• Re-use existing BGP procedures while minimizing the network migration to ease operation
Solution• Let PE1 and PE2 know that they are connected to
the same site using BGP-AD AFI/SAFI, procedures– Add to BGP-AD NLRI an identifier for the Multi-homed Site – Site ID to be the same between peer PEs
• PW infrastructure can be built using either LDP or BGP signaling
• Local Multi-homed PE(s) decide which PE is the designated forwarder for a given site/CE using BGP attributes– LPREF e.g.– VSI ID is the tie-breaker
• If none of the attached CE(s) is elected as designated forwarder for a given VSI, PW status can be used to minimize the BUM traffic in the network
Solution
• Designated PE forwards packets from and to CE
• Non-designated PEs (losing PEs) drop packets from CE as well as from other PEs
• The effect is as if CE was single-homed to just the designated PE
Proposal• BGP AD MH NLRI to carry the VSI-ID
identifying the base VSI and 2 byte Site ID 0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Length (2B) (=14) | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | Route-distinguisher (RD) (8B) | + +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | VSI-ID (HO bits) (2B) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | VSI-ID (LO bits) (2B) | Site ID (2B) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Operation
CE2
VPLS
PE1
PE2
PE3
PE4
Site1LPREF=200
CE3
CE4
CE1
Site2LPREF=200
Site1LPREF=100
Site2LPREF=100
PE1 is elected as designated forwarder
Operation: link failure
CE2
VPLS
PE1
PE2
PE3
PE4
Site1LPREF=200
CE3
CE4
CE1
Site2LPREF=200
Site1LPREF=100
Site2LPREF=100
PE1 withdraws the MH NLRI, send a MAC flushPE2 becomes designated forwarder for CE1
Operation: PE failure
CE2
VPLS
PE1
PE2
PE3
PE4
Site1LPREF=200
CE3
CE4
CE1
Site2LPREF=200
Site1LPREF=100
Site2LPREF=100
PE2 becomes designated forwarder for CE1/CE2 upon BGP neighbor failure detection or upon reception of MH-NLRI
withdraw for CE1/CE2
Operation continued
• RR usage does not impact the operation since the VSI-ID is unique
• No impact in Inter-AS scenario’s since the designated forwarding decisions are local multi-homed PE(s)
• No impact on H-VPLS operation
Next steps
• Minimize BGP flooding through the usage of ORF
• Add startup procedures and procedures for re-configuration
• Do we use this framework for both LDP VPLS and BGP VPLS?