160 likes | 261 Views
Searching for Stability in Interdomain Routing. Rahul Sami (University of Michigan) Michael Schapira (Yale/UC Berkeley) Aviv Zohar (Hebrew University). Border Gateway Protocol (BGP). Akamai. Yahoo!. AT&T. Comcast. Path-vector routing Routing between Autonomous Systems
E N D
Searching for Stability in Interdomain Routing Rahul Sami (University of Michigan) Michael Schapira (Yale/UC Berkeley) Aviv Zohar (Hebrew University)
Border Gateway Protocol (BGP) Akamai Yahoo! AT&T Comcast • Path-vector routing • Routing between Autonomous Systems • ASes can apply routing policies
Convergence/Oscillation Uncoordinated policies can lead to persistent global route oscillations • [Varadhan, Govindan, Estrin] • [Griffin, Wilfong], [Griffin, Shepherd, Wilfong] • Several sufficient conditions for stable convergence [GR01, GGR01,GJR03,FJB05,..] • open question: can a network have two stable solutions, but no oscillation?
Our Results Two stable solutions imply potential BGP oscillations
Our Results • Two stable solutions imply potential BGP oscillations • If preferences satisfy Gao-Rexford constraints • Convergence of n AS network could require Ω(n) timein the wost case • with α-level hierarchy, BGP converges after at most 2α+2 “phases”
BGP model: Routes and Preferences route dest … Prefer AS27 Prefer shorter … AS1 AS3;AS1 AS1 AS27;AS3;AS1 AS1 AS8; AS4;AS1 AS2 AS4;AS2 • Atomic AS/ representative router • Router state: • Available routes to each destination • Route preference rules • Currently selected route • Abstract away export filters, MEDs, etc.
BGP model: Dynamics (for any one destination) j • Each AS i actions: • select best route from available routes • advertise current route to neighbor j • Evolution governed by sequence of action events • Arbitrary (adversarial) timing, with two restrictions: • Fair sequence (no starvation) • Messages not delayed in transit (though may be dropped/lost) i k
State-Transition Graphs * State: profile of all routers’ current routes and beliefs about their available routes Transition: change following route selection or advertisement
State-Transition Graphs * * Zero state State: profile of all routers’ current routes and beliefs about their available routes Transition: change following route selection or advertisement
State-Transition Graphs * * Zero state Stable state(s) State: profile of all routers’ current routes and beliefs about their available routes Transition: change following route selection or advertisement
Main Proof sketch: Regions * • Stable states: blue, red, … • Nonstable states: • blue if all paths lead to blue stable state • red if all paths lead to red stable state • purple otherwise
Proof Sketch: Confluence p a b b a ? a,b : different actions a • Key lemma: from any purple state p, there is a (fair) path to another purple state • Proof: • If all paths to red states, p would be red • cannot have paths to both blue and red state: • => must have path to some purple state p’
Main result: Summary If there are 2 or more stable states, zero state is purple From every purple state, fair path to another purple state Finite number of states=> must cycle sometime => BGP can oscillate on this instance!
Convergence Time • Gao-Rexford conditions • Assume: longest cust-prov chain length is α • Asynchronous model • “Phase”: each router triggered at least once • Result: reach stable solution in at most 2α+2 phases
Discussion & Future Work Main result applies to [GSW] and other models Average case instead of worst-case? Compositional theory for safe policies?
Thank you Questions?