On Using P2P Technology for Decentralized Detection of Service Level Agreement Violations

TL;DR

This paper explores leveraging P2P technology to enhance decentralized detection of SLA violations, aiming to improve scalability, adaptability, and reduce resource consumption in network performance monitoring.

Contribution

It introduces a P2P-based framework for decentralized decision-making and adaptive measurement activation to better detect SLA violations without human intervention.

Findings

P2P principles enable decentralized measurement control.

Adaptive strategies improve detection coverage.

Reduces resource usage compared to traditional methods.

Abstract

Critical networked services enable significant revenue for network operators and, in turn, are regulated by Service Level Agreements (SLAs). In order to ensure SLAs are being met, service levels need to be monitored. One technique for this involves active measurement mechanisms which employ measurement probes along the network to inject synthetic traffic and compute the network performance. However, these mechanisms are expensive in terms of resources consumption. Thus, these mechanisms usually can cover only a fraction of what could be measured, which can lead to SLA violations being missed. Besides that, the definition of this fraction is a practice done by human administrators, which does not scale well and does not adapt to highly dynamic networking patterns. In this article, we examine the potential benefits of using P2P technology to improve the detection of SLA Violations. We first describe the principles of a P2P-based steering of active measurement mechanisms. These principles are characterized by a high degree of decentralized decision making across a network using a self-organizing overlay. In a second step, we present measurement session activation strategies based on these principles. These strategies do not require human intervention, are adaptive to changes in network conditions, and independent of the underlying active measurement technology.