Predict that the sql text will consume high resources

[swetar]

Hi everyone,
Req# Predict that the sql text running in the system will consume high resources.
I wanted to predict this based on the temp space and undo table space used by the query.
I am thinking to use Clustering algo.
Can anyone suggest me, which algo should I use.

Thanks

[DavidHourani]

Are you using MLTK ?

[swetar]

yes , I am using Splunk MLTK.
Is it recommended to do it with Clustering algorithm K-means

[DavidHourani]

yeah that makes sense if you have multiple queries and you wish to classify them, is that what you're trying to achieve ?

[swetar]

Yes there will be multiple queries. I wanted to classify bad sql ( bad sql meaning, sql which are going to consume more than 80% temp or consuming high io)

[DavidHourani]

In that case try BDSCAN could be better than fitting into a set number of clusters.
https://docs.splunk.com/Documentation/MLApp/4.3.0/User/Algorithms#DBSCAN