01-02-2013, 04:40 PM
Extracting Message Types from BlueGene/L’s Logs
ABSTRACT
Event logs consist of several independent lines of text data which contain information about activity on a system. System administrators therefore turn to them frequently as part of daily routine or when they need to detect or diagnose malfunction or misuse. However, event logs can sometimes contain such large amounts of data that manual analysis becomes cumbersome, leading some system administrators to ignore them all together or to miss vital information in them when they do analyze them manually. This fact has made the development of tools and techniques for automatically analyzing the contents of event log file an important area of research. A basic task in automatic analysis of log files is message type or event cluster extraction. Extraction of message types makes it possible to abstract the contents of event logs and facilitates further analysis and the building of computational models. Message type descriptions are the templates on which the individual messages in any event log are built.