DejaVu: Fault Dependency Graphs and Metrics for Online Service Failures

Tsinghua University researchers released datasets for the ESEC/FSE'22 paper on fault localization. The collection includes fault dependency graphs (FDGs), system metrics, and labeled failure ground truths. It supports research into actionable and interpretable diagnosis of recurring failures in online service systems.

Use Cases

• Training fault localization models based on fault dependency graphs (FDGs)
• Evaluating anomaly detection algorithms using labeled failure ground truths
• Analyzing recurring failure patterns in online service systems based on system metrics

Strengths

• Includes labeled ground truth data for failures, enabling supervised learning
• Provides fault dependency graphs (FDGs) for structured system representation
• Associated with a peer-reviewed ESEC/FSE'22 paper, suggesting academic rigor

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

Tsinghua University

Collection Method

Likely collected from online service systems for research purposes.