On the Death of BGP MSN July 8, 2004 Timothy G. Griffin Intel Research, Cambridge UK

Hacking our way to disaster … customer Provider B (Tier 1) Provider A (Tier 1) Provider C (Tier 2) customer sends community that lowers local preference below a provider’s primary backup peering provider/customer

BANG! customer Provider B (Tier 1) Provider A (Tier 1) Provider C (Tier 2) primary backup customer is happy that backup was installed … peering provider/customer

OUCH! customer Provider B (Tier 1) Provider A (Tier 1) Provider C (Tier 2) primary backup One “solution” --- reset BGP session on backup link! This is a stable BGP routing! It’s NOT cisco’s fault! Better --- C should translate its customer depref communities to those of Provider A when re-exporting routes to A (not DT!) peering provider/customer 1/2 Wedgie

REALLY, OUCH! LoadBalancer BELL NET NetNet (Tier 2) HappyPackets (Tier 2) backup primary CIRCUIT NET P1 P2 backup primary ¾ Wedgie

What the heck is going on? There is no guarantee that a BGP configuration has a unique routing solution. –When multiple solutions exist, the (unpredictable) order of updates will determine which one is wins. There is no guarantee that a BGP configuration has any solution! –And checking configurations NP-Complete Complex policies (weights, communities setting preferences, and so on) increase chances of routing anomalies. –… yet this is the current trend! And you have to code your policies in an assembly language that was never designed, but has EVOLVED to allow the required HACKS

Deconstruction BGP TCP MSP TCP Generic Path Vector Specification of routing message semantics MSP = Managed Session Protocol (Veiera, Griffin) SCTP or

Routing Algebras to the Rescue! Network Routing with Path Vector Protocols: Theory and Applications. Joao Luis Sobrinho. SIGCOMM 2003 A node exports to its peers routes learned from its customers A node does not export to a provider a route learned from another provider  crp  ccc  rrr  ppppp  Signature Label c c r p c

Really Hard Bits Tools and techniques for design and development of new Algebras (ongoing work with João Sobrinho) Auto-generation and checking of global constraints –Generalize distrbuted cycle detection algorithms. –Error reporting? Generalize interface between routing protocols and forwarding engine

Just Imagine …. Inventing, implementing, and deploying new routing Algebras without ever going to the IETF or a vendor. Selecting a routing Algebra from a certified library. Working with another Autonomous Systems to develop a routing algebra that is best suited to solving your problems. Having a guarantee that the network is free of anomalies caused by policy interactions…