Balancing by PREFLEX: Congestion Aware Traffic Engineering

Paper type: 
Conference paper
João Taveira Araújo, Richard G. Clegg, Imad Grandi, Miguel Rio and George Pavlou
Networking 2011, Lecture Notes in Computer Science (6641)
There has long been a need for a robust and reliable sys- tem which distributes traffic across multiple paths. In particular such a system must rarely reorder packets, must not require per-flow state, must cope with different paths having different bandwidths and must be self-tuning in a variety of network contexts. PREFLEX, proposed herein, uses estimates of loss rate to balance congestion. This paper describes a method of automatically adjusting how PREFLEX will split traffic in order to balance loss across multiple paths in a variety of network conditions. Equations are derived for the automatic tuning of the time scale and traffic split at a decision point. The algorithms described allow the load balancer to self-tune to the network conditions. The calculations are simple and do not place a large burden on a router which would imple- ment the algorithm. The algorithm is evaluated by simulation using ns-3 and is shown to perform well in a variety of circumstances. The resulting adaptive, end-to-end traffic balancing architecture provides the necessary framework to meet the increasing demands of users while simultaneously offering edge networks more fine-grained control at far shorter timescales.
This paper considers the problem of balancing traffic across network egresses. It achieves a workable solution using a scalable packet market scheme which couples end-hosts controlling their own connection with an overall controller which can select routes appropriately for each flow. The flows are balanced to seek paths which minimise loss.
@incollection{preflex_2011 year={2011}, booktitle={NETWORKING 2011}, volume={6641}, series={Lecture Notes in Computer Science}, title={Balancing by PREFLEX: Congestion Aware Traffic Engineering}, publisher={Springer Berlin Heidelberg}, author={Jo\~{a}o Ara\'{u} jo Taveira and Richard G. Clegg and Imad Grandi, and Miguel Rio and George Pavlou}, pages={135--149} }