Performance evaluation of quality of service aware request placement techniques

Abstract

Web services have become very popular over the past few years. However, users are facing the problem of the perceived QoS. Practical approaches to address this issue are based on service replication at multiple locations and on allocation mechanisms to dispatch requests in such a way that the QoS perceived by the user is improved. In this paper we compare user perceived QoS performance of three dynamic request placement techniques. One technique randomly allocates Web requests while the other two techniques implement a market-based mechanism for QoS negotiation and request allocation. The analysis is conducted via simulation with SIDE using a behavioral model of Web traffic to simulate accurate traffic conditions and determine the limitations or validity of these techniques.