QoS-Aware Service Composition for Work?ow-Based Data-Intensive Applications

摘要

Web service composition can make use of distributed services with various functions to accomplish a specific task. Service binding and composition become keywords to service computing. In the context of workflow management, service binding plays an important part in forming a concrete workflow after an abstract one is submitted. Existing research is devoted to selecting a set of servers to run the corresponding services in an application and achieving better QoS. Due to the data intensity of data flow, the communication between two server nodes may bring significant delay that can not be neglected when considering the performance of the whole application process. In this work, we consider not only the QoS on each server but also the transfer delay of data that affects the total latency. Algorithms are designed to determine a set of service candidates for web service workflow with different structural characteristics. We measure and utilize indispensable parameters to implement these algorithms. Further, we propose Lagrange and ARP algorithms to make comprehensive decisions when taking into account other QoS metrics besides latency.