Optimisation of power consumption of Data centers using a new “resource match” algorithm, which involves characterisation of jobs and machines based on resource requirements and capacity, using techniques of Machine Learning like Hierarchical Clustering and K-means clustering, and OS concepts like first fit, round robin and best fit for scheduling the jobs. The clusters of jobs and machines were matched dynamically with accommodation for future jobs, which optimised the amount of resource consumed by each machine, in turn reducing the consumption of the data center. The research was carried out on the Google Cloud Cluster Trace Dataset.