Anomaly detected data saving location, force historical analysis use more resources (CPU,RAM)

OpenSearch
1

After trying to use Anomaly detection in Opensearch 1.3.6, I still don’t know where the saving location for data points created by Opensearch is and how much storage it takes by using the detector.

Another thing that I noticed is that the History analysis of 13M records took nearly 3h to finish but CPUs and RAM used in this process is very low (compared to normal run without using a detector). So I want to manually change it to use more CPUs and RAM for analyzing faster.

Thank you for reading my topic, hope you have a good weekend !

2

Thanks for your question, I think you can tune these two settings plugins.anomaly_detection.batch_task_piece_size and plugins.anomaly_detection.batch_task_piece_interval_seconds to speed up historical analysis. Refer to Settings - OpenSearch documentation

Sample request to update setting

PUT _cluster/settings
{
  "persistent" : {
    "plugins.anomaly_detection.batch_task_piece_interval_seconds" : 1
  }
}
1 Like
3

Thank you very much! I will give it a shot soon! Hope you have a good day!