Data replication and scheduling in the cloud with optimization assisted work flow management

Data-intensive applications must be run on systems with high-performance processing and enough storage. When compared to conventional distributed systems like the data grid, cloud computing offers these features on a platform that is more flexible, scalable, and inexpensive. Moreover, retrieving data files is crucial to operate these services. Typically, accessing data causes the entire cloud workflow system to experience a bottleneck, thus significantly reducing system performance. Two key strategies that can enhance the efficiency of data-intensive applications are task scheduling and data replication. This research proposes a novel Data replication and scheduling in the cloud. Initially, the workflow management process is performed with 3 phases (1) workflow placement, (2) clustering of tasks, and (3) scheduling and replication. Initially, the workflow placement takes place. Then, the clustering of tasks is performed via an improved K-means algorithm. Finally, the tasks and datasets are replicated during the scheduling and replication phase. Further, the scheduling and replication are performed using the Self Modified Pelican Optimization Algorithm (SM-POA) based on the execution cost, migration cost, storage cost and replication.