Efficient Distributed Algorithm for Scheduling Workload-Aware Jobs on Multi-Clouds

Dynamic distributed algorithm for provisioning of resources has been proposed to support heterogeneous multi-cloud environment. Multi-cloud infrastructure heterogeneity implies the presence of more diverse sets of resources and constraints that aggravate competition among providers. Sigmoidal and logarithmic functions have been used as the utility functions to meet the indicated constraints in the Service Level Agreement (SLA). Spot instances as the elastic tasks can be supported with logarithmic functions while the algorithm always guaranteed sigmoidal functions have the priority over the elastic tasks. The model uses diverse sets of resources scheduled in a multi-clouds environment by the proposed Ranked method in a time window "slice". The paper proposes multi-dimensional self-optimization problem in distributed autonomic computing systems to maximize the revenue and diminish cost of services in the pooled aggregated resources of multi-cloud environment.