The Commodity

Jewell, Josh

doi:10.5040/9781501351143.0022

Cited by 1 publication

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We next explain how DOTE extends to two other central TE objectives: maximizing network throughput [18,22,24,25] (maximum multicommodity flow) and maximum concurrentflow [11,29,48].…”

Section: On Maximum and Concurrent Flowmentioning

confidence: 99%

“…The maximum-multicommodity-flow and maximumconcurrent-flow objectives. In maximum-multicommodity-flow [18,22,24,25], the performance objective L(R , D) is to compute, for a given demand matrix D, a TE configuration R that maximizes the expression L(R , D) = Σ s,t∈V f st (R , D) (the total network throughput). For a TE configuration R and demand matrix D, let α(R , D) denote the maximum value α ∈ [0, 1] for which at least an α-fraction of each D s,t is routed concurrently, i.e., such that for all s,t ∈ V , f st (R , D) ≥ αD s,t .…”

Section: On Maximum and Concurrent Flowmentioning

confidence: 99%

See 1 more Smart Citation

A Deep Learning Perspective on Network Routing

Perry¹,

Frujeri²,

Hoch³

et al. 2023

Preprint

View full text Add to dashboard Cite

Routing is, arguably, the most fundamental task in computer networking, and the most extensively studied one.A key challenge for routing in real-world environments is the need to contend with uncertainty about future traffic demands. We present a new approach to routing under demand uncertainty: tackling this challenge as stochastic optimization, and employing deep learning to learn complex patterns in traffic demands. We show that our method provably converges to the global optimum in well-studied theoretical models of multicommodity flow. We exemplify the practical usefulness of our approach by zooming in on the real-world challenge of traffic engineering (TE) on wide-area networks (WANs). Our extensive empirical evaluation on real-world traffic and network topologies establishes that our approach's TE quality almost matches that of an (infeasible) omniscient oracle, outperforming previously proposed approaches, and also substantially lowers runtimes.

show abstract

“…We next explain how DOTE extends to two other central TE objectives: maximizing network throughput [18,22,24,25] (maximum multicommodity flow) and maximum concurrentflow [11,29,48].…”

Section: On Maximum and Concurrent Flowmentioning

confidence: 99%