Deploying Data-Intensive Service Composition with a Negative Selection Algorithm

With the development of information technology, data on the Internet is growing even faster than Moore's Law. At the age of big data, more and more services are created to deal with big data, which are called data-intensive services. In most cases, multiple data-intensive services are assembled into a service composition to meet complicated requirements. Since the big-data transmission, which is occurred among component services as well as between a service and a data center, has great influence on the overall performance of a composition, deploying those services cannot be considered independently. This paper proposes an optimal deployment method based on a negative selection algorithm for a data-intensive service composition to reduce the cost of the data transmission. When making a deployment schedule, it considers not only the cost of data transmission among component services, but also the load balance of data centers where component services are deployed. It models the deployment problem as a combination optimization problem and extends a negative selection algorithm to get an optimal deployment plan. A series of experiments are carried out to evaluate the performance of the proposed method using different settings as well as to compare with other methods. The results show that the method outperforms others for the problem of data-intensive service composition deployment.