Improving Performance of Internet Services Through Reward-Driven Request Prioritization

Abstract

Meeting client QoS expectations proves to be a difficult task for the providers of modern Internet services, especially when web servers experience overload conditions (steady or transient), which cause increased response times and request rejections, leading to user frustration, lowered usage of the service and reduced revenues. In this paper, we propose a server-side request scheduling mechanism that addresses these problems. Our Reward-Driven Request Prioritization (RDRP) algorithm gives higher execution priority to client web sessions that are likely to bring more service profit (or any other application-specific reward). The method works by predicting future session structure by comparing its requests seen so far with aggregated information about recent client behavior, and using these predictions to preferentially allocate web server resources. Our experiments using the TPC-W benchmark application and CBMG-based web workloads, with an implementation of the RDRP techniques in the JBoss application server, show that RDRP can significantly boost reward attained by the service, while providing better QoS to clients that bring more reward.

Efficient resource management on template-based web servers

“…Instead of scheduling requests based on the estimated service time, they prioritized the requests which are more likely to originate from legitimate clients. Totok and Karamcheti [10] proposed a reward-driven request prioritization mechanism. This mechanism predicts the future structure of web sessions and gives higher execution priority to the requests whose sessions are likely to bring more reward.…”

A business-oriented load dispatching framework for online auction sites

“…For example, once the auction site is under heavy load conditions, instead of rejecting all the requests, it can selectively accept the revenue-generating requests mentioned above. Similar techniques which use priority scheduling for revenue generating transactions were proposed in [22,23].…”

Profit-aware Admission Control for Overload Protection in E-commerce Web Sites

“…However, our ID profiling module does not need to track the internal navigational sequence of a customer session. This is the key difference between our mechanism and RDRP proposed in [13]. RDRP focuses on the inter-request structure within a session, thus relies on the accurate tracking of the internal structure of CBMGs.…”

“…The importance of session integrity has been studied in [4], [5], and session-based admission control (SBAC) mechanisms has been proposed in [5]. More recently, Reward-Driven Request Prioritization (RDRP) mechanism [13] gives higher execution priority to the requests whose sessions are likely to bring more profit. However, these mechanisms focus on inter-request relationship within a session only, and none of them attempt to use the inter-session purchase record of a client for making admission decisions.…”