During the last two decades, parallel computing has drawn attention as an alternative to lessen computational burden in the engineering domain. Parallel computing has also been adopted for meta-heuristic optimization algorithms which generally require large number of functional evaluations because of their random nature of search. However, traditional parallel approaches, which distribute and perform fitness calculations concurrently on the processing units, are not intended to improve the quality of solution but to shorten CPU computation time. In this study, we propose a new parallelization scheme to improve the effectiveness and efficiency of harmony search. Four harmony searches are simultaneously run on the processors in a work station, sharing search information (e.g., a good solution) at the predefined iteration intervals. The proposed parallel HS is demonstrated through the optimization of an engineering planning problem.