Multi-objective Optimization of Ticket Assignment Problem in Large Data Centers

Software or hardware problems in large data centers are usually packaged to be tickets which are assigned to different experts to solve. It is very crucial to design multi-objective ticket scheduling algorithms to maximize the total matching degree and minimize the total flowtime. However, most of existing methods for assignment problems only consider single objective, while some methods optimizing multi-objectives are not for the same objectives of this paper. Meanwhile, exploring effectiveness of existing meta-heuristics for multi-objective optimization could be improved further. In this paper, a multi-objective heuristic algorithm called (GAMOA*) is proposed for ticket scheduling which is the combination of a genetic algorithm (GA) and a multi-objective A* (MOA*). In GAMOA*, ticket scheduling orders are evaluated and improved by GA, while MOA* is applied to find a Pareto set of solutions given an order of tickets effectively and efficiently. Experimental results illustrate that our approach obtains better results than state-of-art algorithms.