query

[donna1up]

[dkapitan]

Goodness me, that is one seriously long query... 😄

[donna1up]

And it runs in 1 minute across 600k patients. This example is just using synthea data.

This runs fairly fast, when you start running views off of views (say that we've flattened and agreed on a standard) it doesn't perform as well. Something to consider, also how it is partitioned if you do that. I dont think you can add indexes in the serverless environment the same as postgres and other sql databases- which would be needed likely if running views off of other views, you'll get timeout errors

[dkapitan]

@donna1up as I was going through the discussion, it came to mind to suggest duckdb as a serverless engine for querying the views. Because it's columnar it performs quite well without indexes. I use it in my workflow by default these days for my projects.

[donna1up]

Thank you, I was actually just learning about duckdb. Thanks for the recommendation!