Towards Logging Noisiness Theory: quality aspects to characterize unwanted log entries
It addresses noise in logging for system monitoring and observability, but it is incremental as it focuses on defining concepts and proposing initial theoretical steps.
This work tackles the problem of noise in log files, which compromises quality by including unwanted information like wrong severity choices and duplicate entries, and proposes initial steps towards a theory called Logging Noisiness to characterize such noise.
Context: Logging tasks track the system's functioning by keeping records of evidence that have been analyzed by monitoring and observability activities. For these activities to be effective, it is necessary to consider the quality of the consumed information. Problem: However, the presence of noise - unwanted information - compromises the log files' quality. The noisiness of a log file can be affected among other things by: (i) the wrong severity log choices, (ii) the production of duplicate entries, (iii) the incompleteness of the information, (iv) the inappropriate format of the entries, (v) the amount of information generated. Objective: This work aims to broadly define the concept of noise in the context of logging, proposing the initial steps of Logging Noisiness, a theory on quality aspects to characterize unwanted log entries.