Characterizing Service Access Patterns under Heterogeneous Clients

Abstract

Service providers usually maintain multiple kinds of service clients, such as desktop Web pages, mobile Web pages, and mobile native apps, to satisfy different user requirements. Understanding how users consume services via these heterogeneous clients is important for service providers to efficiently manage back-end resources. However, little has been known about such facts. To bridge the knowledge gap, in this paper, we conduct the first empirical study on service access patterns via heterogeneous clients. Our study is based on two large-scale real-world datasets of launching events from both app and Web clients of 986 services, involving a sample of 60 million users within 7 days. We analyze the characteristics of service access patterns from both spatial view and temporal view. We explore how users consume online services through different service clients, and to what extent their usage behavior patterns vary when consuming the same service. Our findings provide better understandings for heterogeneous service clients that can facilitate the development, configuration, deployment and maintenance of service back-end.