TY - GEN
T1 - Sweet little lies
T2 - 13th ACM SIGCOMM Workshop on Hot Topics in Networks, HotNets 2014
AU - Vissicchio, Stefano
AU - Vanbever, Laurent
AU - Rexford, Jennifer L.
N1 - Publisher Copyright:
Copyright © 2014 ACM.
PY - 2014/10/27
Y1 - 2014/10/27
N2 - Link-state routing protocols (e.g., OSPF and IS-IS) are widely used because they are scalable, robust, and based on simple abstractions. Unfortunately, these protocols are also relatively inflexible, since they direct all traffic over shortest paths. In contrast, Software Defined Networking (SDN) offers fine-grained control over routing, at the expense of controller overhead, failover latency, and deployment challenges. We argue that future networks can achieve the benefits of both approaches through central control over the distributed route computation. The key idea, which we call Fibbing, is to have the controller trick the routers into seeing a fake topology that is carefully constructed to achieve the desired Forwarding Information Base (FIB). Given an acyclic forwarding graph for each destination, the controller computes an augmented topology with fake nodes (and destinations to announce there) and fake links (and link weights). The controller injects these "lies" into the link-state routing protocol, and the routers simply compute the paths accordingly. The controller can also select an augmented topology that triggers the use of specific backup paths when real links and routers fail. To reduce router load, our Fibbing algorithms compute augmented topologies of minimal size. Our preliminary evaluation on realistic ISP topologies shows that Fibbing works well in practice.
AB - Link-state routing protocols (e.g., OSPF and IS-IS) are widely used because they are scalable, robust, and based on simple abstractions. Unfortunately, these protocols are also relatively inflexible, since they direct all traffic over shortest paths. In contrast, Software Defined Networking (SDN) offers fine-grained control over routing, at the expense of controller overhead, failover latency, and deployment challenges. We argue that future networks can achieve the benefits of both approaches through central control over the distributed route computation. The key idea, which we call Fibbing, is to have the controller trick the routers into seeing a fake topology that is carefully constructed to achieve the desired Forwarding Information Base (FIB). Given an acyclic forwarding graph for each destination, the controller computes an augmented topology with fake nodes (and destinations to announce there) and fake links (and link weights). The controller injects these "lies" into the link-state routing protocol, and the routers simply compute the paths accordingly. The controller can also select an augmented topology that triggers the use of specific backup paths when real links and routers fail. To reduce router load, our Fibbing algorithms compute augmented topologies of minimal size. Our preliminary evaluation on realistic ISP topologies shows that Fibbing works well in practice.
KW - Fibbing
KW - Hybrid SDN
KW - Link-state routing
UR - http://www.scopus.com/inward/record.url?scp=84914680439&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84914680439&partnerID=8YFLogxK
U2 - 10.1145/2670518.2673868
DO - 10.1145/2670518.2673868
M3 - Conference contribution
AN - SCOPUS:84914680439
T3 - Proceedings of the 13th ACM Workshop on Hot Topics in Networks, HotNets 2014
BT - Proceedings of the 13th ACM Workshop on Hot Topics in Networks, HotNets 2014
PB - Association for Computing Machinery
Y2 - 27 October 2014 through 28 October 2014
ER -