TY - GEN
T1 - Analysis of Android Swapping Overhead and Implication of NVM-based Acceleration
AU - Bahn, Hyokyung
AU - Kim, Jisun
N1 - Funding Information:
This work was supported by the National Research Foundation of Korea (NRF) grant fUnded by the Korea government (MSIP) (No. 2019R1A2C1009275) and also by the ICT R&D program of MSIP/IITP (2019-0-00074, developing system software technologies for emerging new memory that adaptively learn workload characteristics).
Publisher Copyright:
© 2020 IEEE.
PY - 2020/12/16
Y1 - 2020/12/16
N2 - Due to the advances in mobile software technologies as well as the rapid diffusion of smart devices, a variety of apps such as online games, social network services, and location-based services emerge every day. Unlike desktop PCs, however, smartphone platforms usually terminate apps without user's agreement if there is not free memory space. This did not incur significant problems in early days, but now it is serious as official works like video conferencing, stock trading, and social broadcasting, are also performed on smartphones. Instead of terminating apps, smartphones can back up the apps to swap storage, but the current smartphones do not make use of it as swapping incurs heavy I/O traffic. Our finding is that the overhead of swapping can be reduced by increasing the memory capacity. However, swapping still incurs performance degradations as the number of apps in execution increases. We show that a small NVM cache can resolve this issue by the workload characterization studies. Specifically, the NVM size necessary for supporting swap without performance degradations is about 1/8 of the main memory capacity if we adopt NVM at the front-end of storage rather than memory layers.
AB - Due to the advances in mobile software technologies as well as the rapid diffusion of smart devices, a variety of apps such as online games, social network services, and location-based services emerge every day. Unlike desktop PCs, however, smartphone platforms usually terminate apps without user's agreement if there is not free memory space. This did not incur significant problems in early days, but now it is serious as official works like video conferencing, stock trading, and social broadcasting, are also performed on smartphones. Instead of terminating apps, smartphones can back up the apps to swap storage, but the current smartphones do not make use of it as swapping incurs heavy I/O traffic. Our finding is that the overhead of swapping can be reduced by increasing the memory capacity. However, swapping still incurs performance degradations as the number of apps in execution increases. We show that a small NVM cache can resolve this issue by the workload characterization studies. Specifically, the NVM size necessary for supporting swap without performance degradations is about 1/8 of the main memory capacity if we adopt NVM at the front-end of storage rather than memory layers.
KW - Android
KW - mobile platform
KW - NVM
KW - smartphone
KW - swapping
UR - http://www.scopus.com/inward/record.url?scp=85105430684&partnerID=8YFLogxK
U2 - 10.1109/CSDE50874.2020.9411538
DO - 10.1109/CSDE50874.2020.9411538
M3 - Conference contribution
AN - SCOPUS:85105430684
T3 - 2020 IEEE Asia-Pacific Conference on Computer Science and Data Engineering, CSDE 2020
BT - 2020 IEEE Asia-Pacific Conference on Computer Science and Data Engineering, CSDE 2020
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 16 December 2020 through 18 December 2020
ER -