Disabling Cache/Optimisations for Benchmarking

Hi Team,

Hopefully this should be an easy question, I am using Dremio to demonstrate to a client its performance on some of their datasets. To get a feel for the raw performance I would like to disable the cache or any optimisations that are similar that could not make the results repeatable.

The table I am querying is an iceberg table, I do not have any reflections on this and the source is ADLS which is configured with ‘Enable local caching when possible’ unchecked.

The queries are being submitted via the python api using arrow flight.

Is there anything else I should considered?

@nikhil.makan Unchecking local cache will give you raw performance, assuming you do not have any reflections created on the dataset

Thanks @balaji.ramaswamy seems like I am on the right track then.