Web Service Flash Crowd Mitigation Using Feedback to Users

摘要

A flash crowd is a situation in which a large number of users access a Web service simultaneously. Conventional methods for managing flash crowds include dropping of excessive requests, preparing of redundant servers, and dynamic provisioning of server resources. It is, however, difficult to use these methods to satisfy all the related requirements such as prevention of system down, avoidance of user abandonment, and efficient use of resources. With a focus on the psychological factor of users, we propose a method for flash crowd mitigation based on the presentation of feedback information. The proposed method was confirmed by experiment to be effective for influencing user behaviors such as abandonment and retransmission of requests. Consequently, we developed a user behavior model and used it to conduct website stress tests using a simulated website. The results further revealed that feedback information could be used to reduce the server load and user abandonment ratio. This paper also suggests the type of Web services for which feedback information-based flash crowd mitigation is most suitable.