Open AccessCharacterizing File Accesses in Android Applications and Caching Implications
Author(s) -
Soojung Lim,
Hyokyung Bahn
Publication year - 2021
Publication title -
ieee access
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.587
H-Index - 127
ISSN - 2169-3536
DOI - 10.1109/access.2021.3125779
Subject(s) - aerospace , bioengineering , communication, networking and broadcast technologies , components, circuits, devices and systems , computing and processing , engineered materials, dielectrics and plasmas , engineering profession , fields, waves and electromagnetics , general topics for engineers , geoscience , nuclear engineering , photonics and electrooptics , power, energy and industry applications , robotics and control systems , signal processing and analysis , transportation
In this paper, we explore Android applications’ file access characteristics, and find out that smartphone file accesses are different from traditional desktop applications in terms of the following aspects. 1) There exist a limited number of hot blocks, which are accessed consistently during the entire execution of an application. 2) Block accesses in Android are highly biased such that the top 20% blocks account for 80% of total accesses. 3) Hot blocks of the top 100 rankings are mostly involved in SQLite. 4) Unlike desktop applications, file accesses in Android applications are write-intensive. 5) In predicting future file accesses in Android applications, frequency is a better estimator than temporal locality. 6) The effect of traditional buffer cache is limited in Android as file I/O in Android has a lot of synchronous writes, which incurs immediate storage flushing. Based on these analyses, this paper presents the implication of buffer cache management in Android. Specifically, we add a small non-volatile write buffer and present how this write buffer can be managed efficiently. Experimental results show that the proposed scheme improves the storage write traffic by an average of 21.7% and a maximum of 48.1% compared to the conventional buffer cache system.