GA-STC: An Optimization Approach for the Placement of Data-intensive Composite Services in the Cloud Environment

Cloud computing provides a great number of services to users and is rapidly being adopted by various industries due to its convenient service options and cost-effective models. Composite services have been developed to address complex application scenarios and their associated challenges. However, the fast-growing amount of data in the application domain puts a huge burden on the communication between the sub-services and makes the placement of the composite services in different servers of the cloud environment a huge challenge. In the existing research, there is no optimization for the large amount of data transmitted between services, which leads to the problem of uneven communication load and long response time. This paper addresses the placement of data-intensive composite services in the cloud. Firstly, the Data-intensive composite Services Placement Problem (DSPP) is modeled considering service type constraints and resource constraints, with the objectives of shortest execution time and communication load balancing. Then we proposed a heuristic algorithm, called GA- STC (Genetic Algorithm with Service Type Clustering), that clusters services type based on the a priori knowledge of service communication in workflows. The algorithm takes into account communication requirements between sub-services and, as a result, deploys services with intensive data transmission on the same server to reduce communication delay and improve operational efficiency. Extensive experimentation was performed to compare GA-STC, a novel placement algorithm with three traditional alternatives: random, First Fit Decreasing (FFD), and Genetic Algorithm (GA). The outcomes demonstrate that, concerning data-intensive composite services placement, GA-STC outperforms the other algorithms significantly in terms of optimization efficacy and efficiency. This research provides crucial guidance for carrying out efficient and effective placement of data-intensive composite services.