Neural network based optimal placement strategy for service components in cloud computing

Many services in cloud environment tend to deploy its components across diverse geographical locations. Latencies between them contribute high in its increased access time. However, it is very difficult to put all components in one location so an alternate solution is to keep them in a nearby location. Further, each user puts its own constraint in order to access the services resulting in more complexity. We addressed this problem in three stages. In the first stage, we used a classic allocation policy to allocate user request to corresponding components. Then we implemented Artificial Neural Network (ANN) algorithm to identify the patterns and further, for different constraints, we propose an incremental assignment approach. We implemented the approach in a customised simulator and found that the observed results are very encouraging with respect to decreasing the access time of service components.