The computational performance of cross-language plagiarism detection system using winnowing algorithm developed at the Electrical Engineering Department, Universitas Indonesia became an issue for real world application. This research is investigating the parallelization of such system implemented on a lab-scale multicore based private cloud platform using OpenStack. Parallelization was done on the portion of the program where the paragraphs of reference documents are processed. The result of execution time of the overall parallelized computation was able to reach speed up of 1.07 to 3.52 times compared to the execution time of the original serial computation. © 2019 IEEE.