An Efficient Method for Service Level Agreement Assessment

On-line end-to-end Service Level Agreement (SLA) monitoring is of key importance nowadays. For this purpose, recent researches have focused on measuring (when possible), or estimating (most of the times) network Quality of Service (QoS) and performance parameters. Up to now, all the proposed solutions have the drawback of requiring a huge amount of resources with low accuracy, generally leading to unscalable systems. We observe, however, that the accurate estimation of network QoS parameters is not necessarily required for SLA assessment. What is required is an efficient and scalable method to directly detect SLA violations. To this end, this paper makes the following contributions. First, we introduce a polynomial-complexity algorithm based on Hausdorff Distance that efficiently detects SLA violations. Second, we propose a Simplified Hausdorff Distance, which provides better accuracy at lower computational cost. Our solution works on simple to measure time series—the Inter-Packet Arrival Times (IPATs) received in our case. The validity of our proposal is confirmed by comparing with perfect knowledge of the QoS status as well as other existing alternatives in a real testbed.