Consider a paged storage hierarchy with at least two levels L1 and L2, where L1 denotes main storage and L2 secondary storage. Suppose that the unit of replacement for L1 is a single page, and that the L2-to-L1 transfer unit, given a page fault, is an integer number of pages. Then, given a suitable replacement policy for L1, increasing the unit of transfer often results in a lower miss ratio at the expense of increased paging traffic. This paper explores the possibility of adaptively varying the L2-to-L1 transfer unit as a function of the reference history of the data to be fetched. Experiments on traces drawn from two large data base systems suggest that such adaptation can result in improved tradeoffs between miss ratios and paging traffic.
Note: The Institute of Electrical and Electronics Engineers, Incorporated is distributing this Article with permission of the International Business Machines Corporation (IBM) who is the exclusive owner. The recipient of this Article may not assign, sublicense, lease, rent or otherwise transfer, reproduce, prepare derivative works, publicly display or perform, or distribute the Article.