Delayed Internet routing convergence

Abstract

This paper examines the latency in Internet path failure, failover and repair due to the convergence properties of interdomain routing. Unlike s w i t c hes in the public telephony n e twork which exhibit failover on the order of milliseconds, our experimental measurements show t h a t i n ter-domain routers in the packet switched Internet may take tens of minutes to reach a consistent view of the network topology after a fault. These delays stem from temporary routing table oscillations formed during the operation of the BGP path selection process on Internet backbone routers. During these periodsofdelayed c onvergence, w e s h o w that end-to-end Internet paths will experience intermittent loss of connectivity, as well as increased packet loss and latency. We present a two-year study of Internet routing convergence through the experimental instrumentation of key portions of the Internet infrastructure, including both passive data collection and fault-injection machines at major Internet exchange points. Based on data from the injection and measurement o f s e v eral hundred thousand inter-domain routing faults, we describe several unexpected properties of convergence and show that the measured upper bound on Internet inter-domain routing convergence delay is an order of magnitude slower than previously thought. Our analysis also shows that the upper theoretic computational bound on the numberofrouter states and control messages exchanged during the process of BGP convergence is factorial with respect to the number of autonomous systems in the Internet. Finally, w e demonstrate that much o f the observed convergence delay s t e m s from speci c router vendor implementation decisions and ambiguity in the BGP speci cation.

OASIS: An Overlay Abstraction for Re-architecting Large Scale Internet Group Services

“…Furthermore, any intermediate peer can react on link failures by passing a packet with unaltered destination prefix to a life neighbor. The latter option on seamless routing resilience happens in contrast to BGP failure convergence times of 2 to 15 minutes [20].…”

Research challenges in QoS routing

“…For instance, in many cases BGP requires tens of minutes to recover from a route or a link failure [56]. Moreover, even though BGP allows an AS to flexibly manage its outbound traffic, it exhibits a scarce degree of control in order to manage and balance how traffic enters an AS across multiple possible paths.…”

A Proposal for Inter-domain QoS Routing Based on Distributed Overlay Entities and QBGP

“…When a violation is detected, the OE in the source AS is capable of reconfiguring onthe-fly its traffic pattern to the remote AS for the affected CoS. Here, the time scale needed to detect and react to a certain problem is very small when compared with the BGP time scale [7]. The end-to-end measurements are based on active AS path probing among peering OEs.…”