.png){kind=icon}
Return
Logfun: An efficient function-Level log management framework for systems implemented with python
.png){kind=icon}
PRE
AI
DOI：10.1016/j.jss.2026.112854.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Logging provides essential execution information for debugging and troubleshooting. Current applications typically integrate logging frameworks to enhance log standardization and unified management. However, existing logging frameworks significantly impact the efficiency of program execution and consume substantial storage resources. Because they lack mechanisms to automatically detect high-frequency or redundant logs and enforce precise dynamic logging controls. During log analysis, it is extraordinarily hard for these frameworks to provide raw log templates and maintain consistent log sequence integrity. Therefore, we propose LogFun, a function-level logging framework that achieves efficient logging and runtime control. It enables efficient log compression and extraction of log templates and variables, as well as automatically identifies high-frequency and redundant logs natively. The experiments show that LogFun decreases the function latency by 23% compared to current logging frameworks. The runtime log controlling decreases latency by 59% and reduces the log volume by 63% in high-load scenarios. LogFun significantly reduces log volume and achieves a compression ratio of 75x to 170x when combined with compression tools. Additionally, LogFun provides precise log templates and variables, reducing the pre-processing overhead associated with log anomaly detection and improving model accuracy.
Keywords:
Logging framework
Log compression
Function as a service
Large language model
Journal
IF:
4.1
Papers:
5.5K
Citations:
8.4K
Organization
Cited Papers
.png){kind=icon}
ShareSave
ShareSaveLog4Perf: suggesting and updating logging locations for web-based systems' performance monitoring
PREAI
ShareSaveShareSave
View more