Multi-level Thresholding Partitioning Algorithm for Graph Processing in Cloud Computing

Graphs are an effective way for data representation, and generally, the graph data size is enormous. Graphs are linked with various revenue-generating applications such as social media, online retail, drug discovery, clinical trials, and businesses. Partitioning is a mechanism used for processing the vast graph data effectively. Graph partitioning efficiently minimizes the energy consumption and computational complexity caused while processing the extensive interlinked data. In this work, node priority and threshold-based graph partitioning algorithms are proposed for achieving energy-efficient graph processing. Most graph partition algorithms promptly minimize energy consumption, but the node priority-based graph partitioning algorithm also delivers enhanced results in the aspect of execution time. In this research work, the power consumption is measured through online power estimation tools. To calculate the energy consumption, in this work, five popularly known graph applications are deployed. The power consumption is categorized into static and dynamic power consumption, estimated using four benchmarked graph datasets. The result analysis includes the energy and performance cost according to the processor. The obtained results show that the proposed runtime is enhanced effectively than the existing works, thus achieving the motto of this research work in the aspect of minimizing the overall energy percentage.