Static Versus Adaptive Paging: A Simulation-Based Comparison of Windows-Style and Linux Memory Management Under Realistic Workloads

Abstract

Paging strategies are essential for managing memory and preventing system bottlenecks under high workloads. This study presents a comparative performance analysis between Windows-style static allocation and a Linux-hypothesis adaptive management model. The research employs a dual-methodology approach, combining a Java-based simulation of FIFO, LRU, and Optimal algorithms with a theoretical model of Linux’s active-inactive list logic. These models were evaluated across four realistic memory access patterns: Random, Browser, File Access, and Video Access. Quantitative results from the Java simulation show that while the Optimal algorithm achieves 100% theoretical efficiency as a benchmark, the LRU algorithm demonstrates an efficiency of approximately 80% due to its use of temporal locality. In contrast, the FIFO algorithm shows only 55% efficiency, leading to higher disk I/O and system bottlenecks. The study further explores how Linux’s adaptive approach can theoretically minimize these bottlenecks by dynamically managing memory frames. This research concludes that while standard algorithms provide a foundation for performance, the rigid limits of static allocation necessitate a shift toward more flexible management strategies to maintain system stability.

Keywords: Memory Management, Locality of Reference, Page Faults, Windows vs Linux, Thrashing, Page Replacement Algorithms, LRU Approximation, Performance Bottlenecks, Paging, Adaptive Paging.