Multivariate Log-based Anomaly Detection for Distributed Database
The dataset download link:
- Single2Single: https://zenodo.org/records/11496301/files/Single2Single.tar.gz
- Single2Multi: https://zenodo.org/records/11496255/files/Single2Multi.tar.gz
- Multi2Single: https://zenodo.org/records/11483841/files/Multi2Single.tar.gz
- Multi2Multi: https://zenodo.org/records/11468477/files/Multi2Multi.tar.gz
This dataset is mainly designed for cluster anomaly detection:
| No. | Anomaly | Cause Type | Description |
|---|---|---|---|
| 1 | CPU Saturation | System | The CPU computing resources exhaust. |
| 2 | IO Saturation | System | The I/O bandwidth is heavily occupied. |
| 3 | Memory Saturation | System | Insufficient memory resources. |
| 4 | Network Bandwidth Limited | System | The network bandwidth between nodes is limited. |
| 5 | Network Partition Arise | System | Network partition occurs between nodes. |
| 6 | Machine Down | System | One server goes down when the applications are running. |
| 7 | Accompanying Slow Query | Database | Excessive query load. |
| 8 | Export Operations | Database | Backing up data to external source. |
| 9 | Import Operations | Database | Importing data from external source. |
| 10 | Resource-Intensive Compaction | Database | Compaction tasks consume a substantial amount of system resources. |
| 11 | Overly Frequent Disk Flushes | Database | The low interval of flush operations leads to frequent disk writes. |