Panel App which filters just 5M rows of data using dask is very slow - Why?

I have an application which should Read raw data (ultimately, 100’s of millions of rows) Filter data based on widgets Draw graphs I’m evaluating Holoviz for the purpose. Now I have created an application which uses Dask, but it’s really, really slow. I’ve tried to reproduce the slowness in this MWE app with a dummy data set. Code for MWE app: from __future__ import annotations import sys import dask.dataframe as dd import hvplot.dask # noqa import panel as pn @pn.cache def get_data() -> ...