How the 'Transform API' Works


I was check
"search > aggregation (size=0)" and "transform > pivot > aggregation (_preview)"
I checked that the transform side has much fewer tooks.
(Same aggs, small size set for simple comparison -> There was a noticeable difference.)

How did you get these results?
May I know how the transform API works in more detail than refer?
Do you think there are related documents?

Thank you!


transform uses Composite aggregation underneath, wrapped in a search call. There shouldn't be a significant performance difference between a transform preview call or a search. Both calls get executed with the same priority and use the same threadpool.

The documentation has more information and also explains some inner details like when to use transform vs. something else, checkpointing ans scale considerations.

1 Like

Thank you for your exact answer! @Hendrik_Muhs

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.