TY - GEN
T1 - Dynamic merging/splitting for better responsiveness in mobile devices
AU - Wu, Chao
AU - Ji, Cheng
AU - Shi, Liang
AU - Xue, Chun Jason
AU - Huang, Bin
AU - Wang, Yuangang
N1 - Publisher Copyright:
© 2016 IEEE.
PY - 2016/8/17
Y1 - 2016/8/17
N2 - Mobile devices are playing important roles in people's daily life. For mobile devices, system responsiveness is of great importance to keep an impressive user experience. Although new technologies have been developed to improve system responsiveness, mobile devices still suffer from various unpredictable delay. Recently, our evaluation reveals that improper merging operations happen in I/O scheduler layer is an important reason leads to intolerable system latency. In order to mitigate system latency caused by overlarge merged requests, a novel scheme called dynamic merging/splitting based I/O scheduling scheme is proposed. Adding a new unit called "SU" in I/O scheduler, overlarge merged requests are split and re-merged to improve responsiveness of these requests. Evaluation results show that the proposed scheme can effectively improve system responsiveness of android devices. By boosting 20.38% synchronous requests, the latency of small requests are reduced by 35.98%, and large requests are reduced by 48.47% on average. What's more, the worst case latency of synchronous requests are reduced over 90% after optimization.
AB - Mobile devices are playing important roles in people's daily life. For mobile devices, system responsiveness is of great importance to keep an impressive user experience. Although new technologies have been developed to improve system responsiveness, mobile devices still suffer from various unpredictable delay. Recently, our evaluation reveals that improper merging operations happen in I/O scheduler layer is an important reason leads to intolerable system latency. In order to mitigate system latency caused by overlarge merged requests, a novel scheme called dynamic merging/splitting based I/O scheduling scheme is proposed. Adding a new unit called "SU" in I/O scheduler, overlarge merged requests are split and re-merged to improve responsiveness of these requests. Evaluation results show that the proposed scheme can effectively improve system responsiveness of android devices. By boosting 20.38% synchronous requests, the latency of small requests are reduced by 35.98%, and large requests are reduced by 48.47% on average. What's more, the worst case latency of synchronous requests are reduced over 90% after optimization.
UR - https://www.scopus.com/pages/publications/84986562980
U2 - 10.1109/NVMSA.2016.7547185
DO - 10.1109/NVMSA.2016.7547185
M3 - 会议稿件
AN - SCOPUS:84986562980
T3 - 2016 5th Non-Volatile Memory Systems and Applications Symposium, NVMSA 2016
BT - 2016 5th Non-Volatile Memory Systems and Applications Symposium, NVMSA 2016
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 5th Non-Volatile Memory Systems and Applications Symposium, NVMSA 2016
Y2 - 17 August 2016 through 19 August 2016
ER -