Logflow
- Project
- 16037 PAPUD
- Type
- New standard
- Description
Anomaly detection and analysis for system logs in one tool.
- Contact
- Cornel Crisan (Bull)
- Cornel.crisan@atos.net
- Technical features
Input(s):
- System Logs in a flat text file
Main feature(s):
- Detect anomaly from system logs
- Calculate correlation between logs (e.g. a log in anomaly with the previous logs)
Output(s):
- Anomaly indicator
- Graph displaying
- Correlation between logs
- Integration constraints
There are two distinct tools developed during the project:
- Anomaly detection
- Log correlation calculation The two tools can be used independently or can be combined to detect an anomaly and analyze the logs correlation tree for this anomaly. The tools were developed based on the data provided by a Bull HPC system. These data are confidential so we released tools with example data from existing public dataset. Therefore, results of released tools can be differt from original tool optimized for the HPC confidential dataset. Tools have to be modified in order to work with other dataset than those presented in the released tools and may be finetuned to reach good results with other dataset.
- Targeted customer(s)
Datacenter provider or system administrators who want to detect and analyze anomalies in a large amount of system logs.
- Conditions for reuse
Open-source free license tools under Apache 2.0 license. Reuse of tools is authorized under this license terms.
- Confidentiality
- Public
- Publication date
- 26-10-2020
- Involved partners
- Bull (FRA)
- Université de Lorraine (FRA)