Tomography Based Learning for Load Distribution Through Opaque Networks

Shenghe Xu,Murali Kodialam,T V Lakshman,Shivendra S Panwar

doi:10.1109/ojcoms.2021.3068222

Shenghe Xu, Murali Kodialam + Show 2 more

Open Access

https://doi.org/10.1109/ojcoms.2021.3068222

Copy DOI

Abstract

Applications such as virtual reality and online gaming require low delays for acceptable user experience. A key task for over-the-top (OTT) service providers who provide these applications is sending traffic through the networks to minimize delays. OTT traffic is typically generated from multiple data centers which are multi-homed to several network ingresses. However, information about the path characteristics of the underlying network from the ingresses to destinations is not explicitly available to OTT services. These can only be inferred from external probing. In this paper, we combine network tomography with machine learning to minimize delays. We consider this problem in a general setting where traffic sources can choose a set of ingresses through which their traffic enter a black box network. The problem in this setting can be viewed as a reinforcement learning problem with strict linear constraints on a continuous action space. Key technical challenges to solving this problem include the high dimensionality of the problem and handling constraints that are intrinsic to networks. Evaluation results show that our methods achieve up to 60% delay reductions in comparison to standard heuristics. Moreover, the methods we develop can be used in a centralized manner or in a distributed manner by multiple independent agents.

Highlights

R ECENT emerging applications including virtual reality, online or cloud gaming require low delay for acceptable user experience [1], [2]
TWO REPRESENTATIVE PROBLEMS Though the idea of tomography based learning is applicable to several networking scenarios, this paper focuses on two applications, Egress Picking and Traffic Engineering using Segment Routing, to illustrate the applicability of the method
The traffic sources do not use this information in their load distribution decisions, since from their perspective, the network to which they are sending traffic is a black box, and the only usable information about the network is that inferred from external probing

Summary

Introduction

R ECENT emerging applications including virtual reality, online or cloud gaming require low delay for acceptable user experience [1], [2]. Minimizing delay by optimizing load distribution through underlying networks is an important task for providers of these services. Since these services are often “over-the-top” services, the providers do not have full knowledge of the underlying networks and have to make load distribution decisions based purely on inference of the network characteristics from edge-based observations. Inferring network characteristics from external observations, called “network tomography”, has been extensively studied. Our interest is in “active tomography” where probes from the network periphery are used to infer internal network characteristics [6]–[9]

Objectives

Results

Conclusion