Survey on Online Log Parsers

Abstract

Technological dependence is growing in leaps and bounds as days progress. As a result, software applications are required to be up and running at all times without fail. The health and safety of these applications need to be monitored regularly by theuse of constant logging of any faults that occur at their runtime executions. Log analysis techniques are applied to recorded logsto obtain a better overview of how to handle failures and health deterioration. Before these algorithms can be utilized in practice, the raw unstructured logs need to be converted into structured log events. This process is performed by log parsers, which are accessible in two different modes – offline and online. While offline log parsers have a predefined knowledge base containing templates and conversion rules, online log parsers learn new templates on the job. This paper focuses on surveying and creating a comparative study on online log parses by analysing the type of technique used, efficiency and accuracy of the parser on a given dataset, time complexity, and their effectiveness in motivating applications.