Frequent itemset mining and association rule generation are common topics in mining massive datasets or big data, the purpose of which is to discover itemsets bought together with a significant proportion. Traditional algorithms are mostly implemented and evaluated as single-machine in-memory programs that are incapable of large-scale systems. In the study, the authors contrast the performance of two algorithms for frequent itemset mining, including A-Priori and Park-Chen-Yu (PCY), in the Hadoop Distribution File System (HDFS) using the PySpark framework. Empirical results in practical datasets emphasize the remarkable performance of PCY thanks to eliminating non-frequent pairs using buckets of candidate ones.