QoS-Based Service Selection Method for Big Data Service Composition

Different from the traditional web services, the big data services' execution duration vary from the input data volume, so the traditional Quality of Service (QoS) analysis model for traditional web services cannot be directly applied to big data services. Additionally, since many big data or web services provide overlapping or identical functionality, albeit with the description of different QoS, some methods should be taken to determine which service are to participate in a given composite service. To address the problems above, this paper proposes an expanded edition of QoS-based analysis model: (1) Use the linear regression model to estimate the execution duration of the big data services to support the QoS model. (2)Set the weight of QoS using AHP analysis. (3) Improve the service selection algorithm based on backtracking method and validate its effectiveness, which is proved more high-performance than the original backtracking method and Integer Programming method.