Relative Interval Analysis of Paging Algorithms on Access Graphs

TL;DR

This paper applies relative interval analysis to compare paging algorithms like LRU, FIFO, FWF, and FAR on various access graph structures, revealing nuanced performance differences and solving an open problem.

Contribution

It introduces the use of relative interval analysis with access graphs for paging algorithms and provides tight bounds on LRU versus FIFO performance.

Findings

LRU outperforms FIFO on paths

LRU performs worse on stars, cycles, and complete graphs

Tight bounds established for LRU and FIFO relationship

Abstract

Access graphs, which have been used previously in connection with competitive analysis and relative worst order analysis to model locality of reference in paging, are considered in connection with relative interval analysis. The algorithms LRU, FIFO, FWF, and FAR are compared using the path, star, and cycle access graphs. In this model, some of the expected results are obtained. However, although LRU is found to be strictly better than FIFO on paths, it has worse performance on stars, cycles, and complete graphs, in this model. We solve an open question from [Dorrigiv, Lopez-Ortiz, Munro, 2009], obtaining tight bounds on the relationship between LRU and FIFO with relative interval analysis.