A Time-Efficient Solution to the General Resource Placement Problem in Cloud

Abstract EN

Cloud based large-scale online services are faced withregionally distributed stochastic demands for various resources.

With multiple regional cloud data centers, a crucial problem thatneeds to be settled is how to properly place resources to satisfymassive stochastic demands from many different regions.

For thegeneral stochastic demands oriented cross region resource placementproblem, the time complexity of existing optimal algorithmis linear to total amount of resources and thus may be inefficientwhen dealing with a large number of resources.

To end this, wepropose an efficient algorithm, named discrete function basedunbound resource placement (D-URP).

Experiments show thatin scenarios with general settings, D-URP can averagely achieveat least 97% revenue of optimal solution, with reducing timeby three orders of magnitude.

Moreover, due to the generalityof problem setting, it can be extended to get efficient solutionfor a broad range of similar problems under various scenarioswith different constraints.

Therefore, D-URP can be used asan effective supplement to existing algorithm under time-tensescheduling scenarios with large number of resources.