QoS-aware service replication

Abstract

Service composition is a useful technique to assemble light, independent services to meet the complicated and dynamic requirements. Previous research has addressed the quality-of-service (QoS) aware composition path selection problem. However, as the requests growing, the selected service composition path may violate the QoS requirements. In this case, more service replicas should be deployed on suitable nodes to improve the QoS. But which service component should be selected and where these service replicas should be deployed is a challenge. In this paper, we make a deep study on this service replication problem. We give a detailed description of service replication triggering time. And then, we propose LDCS (Longest Delay Service Component Selection) to select the bottle-neck service component by evaluating the real-time performance of all these components. Finally, we employ MACP (Maximum Available Capacity Path) algorithm to select a suitable node to deploy this service replica. Simulation results approve that our approach is effective and efficient.