Analysis on the Scheduling Problem in Transparent Computing

Transparent computing has received increasing attention recently. As a variation and implementation of cloud computing, scheduling is destined to be one of the most hot topics for performance optimization. In this paper, we focus on a typical transparent computing platform, where a number of clients are connected with a server cluster in a gigabit LAN. Due to the NP-completeness of the scheduling problem on parallel processors, the optimal scheduling solution of TC can not be achieved in polynomial time. To find an efficient scheduling algorithm, We present two approximation algorithms and analysed their upper bounds theoretically. Though the analytic results are less than satisfactory, the practical performances of the proposed algorithms are demonstrated to be acceptable by extensive simulations.