A Toolkit to Analyze the Task Allocation Strategies for Large Dataset in Geo-Distributed Servers

The evolving technological improvements on data handling with the geo-distributed datacenters pave the way for the cost-effective data maneuvres. The geo-distributed datacenter-based system planning poses an over whelming challenge in aggregating the appropriate data, distributing the server intensive computation, and setting up the seamless communication infrastructure. The datacenter providers have been striving to reduce the operational expenditure with different measures. The three segments, i.e., task assignment, data placement, and data movement essentially influence the operational cost of datacenters. We considered the cost minimization issue by streamlining of these three segments for advantageous benefits in geo-distributed servers. To delineate the endeavor of choosing the plausibility of the appropriate task allocation strategy, we suggested MATLAB-based toolkit that compares a 2D Markovian chain-based model and a MILP-based task allocation model. A mixed nonlinear programming constructs were linearized for the cost minimization. A MATLAB-based toolkit was proposed to explore the different task allocation strategies and their impact on the operational cost, communication cost, and the overall server cost. The existing datacenter simulation tools are having rigid and fixed functionality whereas the proposed toolkit offers different possibilities for the analysis and visualization of datacenter-related operations.