Load Balancing and Thermal-Aware in Geo-Distributed Cloud Data Centers Based on Vlans

Mustafa I. Khaleel
2018 Science Journal of University of Zakho  
Power consumption in datacenters has become an emerging concern for the cloud providers. This poses enormous challenges for the programmers to motivate new paradigms to enhance the efficiency of cloud resources through designing innovative energy-aware algorithms. However, balancing the weights over geographically dispersed datacenters has been shown to be essential in decreasing the temperature consumption per datacenter. In this paper, we have formulated a load balancing paradigm to exploit
more » ... e idea of scheduling scientific workflows over distributed cloud resources to make system outcome more efficient. The proposed heuristic works based on three constraints. First, initiating cloud resource locality for tenants and calculating the shortest distance in order to direct module applications to the closet resources and conserving more bandwidth cost. Second, selecting the most temperature aware datacenters based on geographical climate to maintain electricity cost for the providers. Third, running multiple datacenters within the same geographical location instead of housing the entire workloads in a single datacenter. This allows providers to take a tremendous advantage of sustaining the system from degradation or even unpredictable failure which in turn will frustrate the tenants. Furthermore, applications are formulated as Directed Acyclic Graph (DAG)-structured workflow. For the underlying cloud hardware, our model groups the cloud servers to communicate as if they were in the same physical location. Additionally, both modes, on-demand and reservation, are supported in our algorithm. Finally, the simulation showed that our method was able to enhance the utilization rates about 67% compared to the baseline model.
doi:10.25271/sjuoz.2018.6.3.515 fatcat:qpdfarh55rfe3ofab3ynwuinzu