Vichalana Anomaly Benchmark

This repository contains a time-series, multivariate dataset which can be used in performance anomaly detection in API Gateways. Vichalana Anomaly Benchmark can be used for evaluating anomaly detection algorithms. It comprises with 25 features and total of ~50,000 data points. This dataset is generated using a industry standard setup of API Gateway.

Our paper can be found at:

Anomaly Detection in High-Performance API Gateways

If you use this dataset, please cite:

@INPROCEEDINGS{9188100,
  author =    {Geethika, Deshani and Jayasinghe, Malith and Gunarathne, Yasas and Gamage, Thilina Ashen and Jayathilaka, Sudaraka and Ranathunga, Surangika and Perera, Srinath},
  booktitle=  {2019 International Conference on High Performance Computing & Simulation (HPCS)}, 
  title=      {Anomaly Detection in High-Performance API Gateways}, 
  year=       {2019},
  volume=     {},
  number=     {},
  pages=      {995-1001},
  doi=        {10.1109/HPCS48598.2019.9188100}}

Dataset Details

Merged anomaly dataset can be found in /combined_dataset The individual parts of the normal dataset can be found in /normal_datasets The anomalous dataset can be found in /anomalous_datasets

Details of anomalous dataset are given below

1. Scenario 01: High CPU usage due to mediation (Type 1)
2. Scenario 02: High Memory usage due to mediation (Type 2)
3. Scenario 03: High disk I/O due to mediation (Type 3)
4. Scenario 04: Increased load (numbers of users) (Type 4)
5. Scenario 05: Increased load (throughput) (Type 4)
6. Scenario 06: Long response time in back-end services (Type 5)
7. Scenario 07: Increased message size (Type 6)
8. Scenario 08: Failure in back-end services (Type 7)
9. Scenario 09: Increased load (throughput) (Type 4)
10. Scenario 10: Increased message size (Type 6)