Association rule mining is one of important data mining problems. In this paper, a framework for efficiently calculating frequent itemsets in voluminous data is presented. The algorithm FIT [LR] is a practical implemention of the framework. A theoretical comparison between FIT and Eclat [ZPOW] is also explored. The analysis asserts that the performance of FIT is much more efficient than that of Eclat. Experimental results confirmed the assertion with data from [AS].