Faster Link-State IGP Convergence and Improved Network Scalability and Stability
LCN '01 Proceedings of the 26th Annual IEEE Conference on Local Computer Networks
Faster Link-State IGP Convergence and Improved Network Scalability and Stability
LCN '01 Proceedings of the 26th Annual IEEE Conference on Local Computer Networks
Traffic Engineering and QoS Optimization of Integrated Voice & Data Networks
Traffic Engineering and QoS Optimization of Integrated Voice & Data Networks
| Hi-index | 0.00 |
It is desirable to have fast detection of failuresfollowed by fast restoration and recovery in datanetworks (IP,ATM,Frame relay,etc.) using link-stateInterior Gateway Protocols (IGP).This requires fasterhello exchanges (unless one relies on link level detectionwhich is not always possible), fast flooding, and morefrequent Shortest-Path-First calculations.However,through analysis and simulation we show that thiscompromises the scalability and stability of the network.This is mainly because Hello packets received at arouter are indistinguishable from other packets and mayexperience long queueing delays during a sudden burstof many LSA (Link-State Advertisement)updates andcause the associated link to be declared down.At a laterinstant the link would recover.Both the failure andrecovery generate fresh LSAs, in effect developing afeedback loop to sustain the LSA storm.We suggest thatHello and potentially some other IGP packets, such asacknowledgments, be marked explicitly so that they mabe distinguished and queued separately from other IGPand data packets at the line rate and provided prioritizedtreatment.This allows significant reduction in IGPconvergence time while maintaining network stability.