Dynamic Service Placement for Mobile Micro Micro-Clouds Clouds with Predicted Future Costs
Abstract: Mobile micro-clouds clouds are promising for enabling performance performance--critical cloud applications. However, one challenge therein is the dynamics at the network edge. In this paper, we study how to place service instances to cope with these dynamics, where multiple u users sers and service instances coexist in the system. Our goal is to find the optimal placement (configuration) of instances to minimize the average cost overtime, leveraging the ability of predicting future cost parameters with known accuracy. We first propos propose e an offline algorithm that solves for the optimal configuration in a specific look look-ahead time-window. window. Then, we propose an online approximation algorithm with polynomial time time-complexity complexity to find the placement in real-time time whenever an instance arrives. We an analytically alytically show that the online algorithm is 0(1) 0(1)-competitive competitive for a broad family of cost functions. Afterwards, the impact of prediction errors is considered and a method for finding the optimal look-ahead ahead window size is proposed, which minimizes an upper bound of the average actual cost. The effectiveness of the proposed approach is evaluated by simulations with both synthetic and real real-world world (San Francisco taxi) usermobility traces. The theoretical methodology used in this paper can potentially be applied to a larger class of dynamic resource allocation problems.