Friday, September 04, 2009

Workshop on the Analysis of System Logs (WASL) 2009


Just reposting the announcement for a fun workshop on logs.

Workshop on the Analysis of System Logs (WASL) 2009

Date: October 14, 2009
Location:  Big Sky, MT (at SOSP conference)

“System logs contain a wide variety of information about system status and health,
including events from various applications, daemons and drivers, as well as sampled
information such as resource utilization statistics. As such, these logs represent a
rich source of information for the analysis and diagnosis of system problems and
prediction of future system events. However, their lack of organization and the general lack of semantic consistency between information from various software and hardware  vendors means that most of this information content is wasted. Indeed, today's  most popular log analysis technique is to use regular expressions to either detect  events of interest or to filter the log so that a human operator can examine it manually.  Clearly, this captures only a fraction of the information available in these logs and does not scale to the large systems common in business and supercomputing environments. 

This workshop will focus on novel techniques for extracting operationally useful
information from existing logs and methods to improve the information content of future  logs.” (quoted from WASL site

Workshop Program

Session 1: Log Analysis Tools
      - "Extracting Message Types from BlueGene/L's Logs", A. Makanju, A. Zincir-Heywood, and E. Milios
       - "Incremental Learning of System Log Formats", K. Zhu, K. Fisher, and D. Walker
       - "Visual and Algorithmic Tooling for System Trace Analysis: A Case Study", W. De Pauw and S. Heisig

Session 2: Analyzing System Logs
       - "Mining Dependency in Distributed Systems through Unstructured Logs Analysis", J. Lou, Q. Fu, Y. Wang, and J. Li
       - "A Bayesian Network Approach to Modeling IT Service Availability using System Logs", R. Zhang, E. Cope, L. Huesler, and F. Cheng
       - "Endpoint Identification Using System Logs", S. Melvin

Session 3: Group Discussion on Current State of the Art
       - Tips and tricks in current use.
       - Gaps and challenges in current techniques.
       - Vision and steps for the future.

Session 4: Panel on Future Research Agenda
       - What are the most difficult problems with logging, in the real world?
       - How to make academia-industry interactions more productive?  [A.C. – will probably be very fun!]
       - How to extract meaningful information from logs?
       - How to improve system management?

More info and sign-up here.

Dr Anton Chuvakin