What Supercomputers Say: A Study of Five System Logs

By Adam J. Oliner, Jon Stearley. Dependable Systems and Networks, 2007

698 citations recorded by the source corpus.

Subfields: Computer Networks and Communications, Management Science and Operations Research

View Paper: https://www.semanticscholar.org/paper/01b5c01835a57f63c250b4eed923b7f736707624

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

1. Hongyu Zhang
2. Van-Hoang Le
3. Michael R. Lyu
4. Zibin Zheng
5. Pinjia He
6. Jieming Zhu
7. Shilin He
8. Pradip Bose

In Management Science and Operations Research

1. Michael R. Lyu
2. Jieming Zhu
3. Pinjia He
4. Van-Hoang Le
5. Hongyu Zhang
6. Zibin Zheng
7. Shilin He
8. Dan Pei

Citations in the corpus

• Drain: An Online Log Parsing Approach with Fixed Depth Tree (2017, methodology): “Specifically, BGL is a log data set collected by Lawrence Livermore National Labs (LLNL) from BlueGene/L supercomputer system [21].”
• Experience Report: System Log Analysis for Anomaly Detection (2016, methodology): “BGLdata contain 4,747,963 log messages, which were recorded by the BlueGene/L supercomputer system at Lawrence Livermore National Labs (LLNL) [36].”
• Addressing failures in exascale computing (2014, background): “A study of major DOE supercomputers in 2004–2006 showed that about 65% of failures could be attributed to software (Oliner and Stearley, 2007), whereas a study of failures in 2012 on Intrepid, the BG/P system at Argonne National Laboratory, showed that less than 16% of job crashes were due to hardwa”
• Toward Exascale Resilience (2009, background): “Several analyses have been made concerning the root cause behind failures (Lu, 2005; Oliner and Stearley, 2007; Schroeder and Gibson, 2007).”
• Tools and Benchmarks for Automated Log Parsing (2018, methodology): “With close collaborations with our industrial partners, as well as some pioneer researchers (authors from [10], [18], [30]), we collect a large set of logs (over 77GB in total) produced by 16 different systems spanning distributed systems, supercomputers, operating systems, mobile systems, server ap”
• LogAnomaly: Unsupervised Detection of Sequential and Quantitative Anomalies in Unstructured Logs (2019, methodology): “Datasets We conduct experiments over the BGL dataset [Oliner and Stearley, 2007] and the HDFS dataset [Xu et al., 2009], as listed in Table 2.”
• Log-based Anomaly Detection Without Log Parsing (2021, methodology): “The datasets are collected between 2004 and 2006 from real-world supercomputing systems [33] and consist of 14,672,653 log messages in total, among which 353,394 log messages are manually labeled as anomalies.”