Embedded DBMS Design for In-Vehicle Information Management

被引:2
作者
Oh, Joontaek [1 ]
Won, Youjip [1 ]
机构
[1] Hanyang Univ, Seoul, South Korea
来源
2018 7TH IEEE NON-VOLATILE MEMORY SYSTEMS AND APPLICATIONS SYMPOSIUM (NVMSA 2018) | 2018年
关键词
D O I
10.1109/NVMSA.2018.00028
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Embedded devices have built-in software, which results in them having to include storage for software. Embedded devices use flash storage since it should be strong against physical impact. However, flash storage has a limited lifespan, which is equivalent to the lifespan of embedded device. Relatively inexpensive embedded devices can tolerate such limitation, but in the case of high cost devices like smartcars, such ephemeral lifespan is unacceptable. In this paper, we improve the lifespan of smartcars by improving the lifespan of their flash storage, by minimizing the write volume of SQLite. SQLite that is serverless DBMS that is mainly used in embedded devices stores the log in a separated file, journal file, at commit time. This is a logging. Logging doubles the write volume because the SQLite writes data to both the log file and the database file. Moreover, when using a journaling filesystem such as EXT4, the write volume is amplified by the journaling of journal anomaly. It leads to more program/erase operations in flash storage. It short the lifespan of flash storage. This anomaly leads to the shortened lifespan of its embedded device. If the embedded device is a smartcar and the lifespan is reduced down to five years due to write amplification, we would have to pay extravagant cost every five years. We need to optimize for long-term use of smartcars before using them. There have been many efforts to reduce the write volume of SQLite. This paper examines how smartcars can improve their lifespan when a previous study to reduce the write volume of SQLite is applied to smartcars. [GRAPHICS] .
引用
收藏
页码:111 / 112
页数:2
相关论文
共 4 条
[1]  
Jeong C, 2014, IEEE INT SYMP INFO, P61, DOI 10.1109/ISIT.2014.6874795
[2]  
Kim Jaegeuk, F2FS: support atomic_write feature for database
[3]  
Lee W., 2015, WALDIO ELIMINATING F
[4]  
Park Changhyun, 2017, ACM SPEC INT C PERF, P257