1

Issue Summary

I have a dashboard with nine widgets and they are all scattered plot running the same query which return a data frame which has about 14500 rows and 10 columns. The query takes 500 ms to run on the database but the dashboard is taking around 4 minutes to load.

I tried writing a separate query for each widget, so each query will only return the data needed for that particular plot which is about 14500 rows and 4 columns. I was able to drop the total dashboard loading time to about 2 minutes.

Any suggestion on how to further optimize the loading time for the dashboard?
Thank you very much.

Technical details:

  • Redash Version: 8.0.0
  • Browser/OS: Chrome (89.0.4389.90)
  • How did you install Redash: Self-hoseted Redash on AWS
2

Open your browser’s network inspector when you load the dashboard. Where is the bottleneck?

Does it take several minutes for the data to return from your database? Or is it just the browser taking a long time to draw the visualization?

Side note: Redash isn’t built to display that many data points on a 3" wide widget on your screen. I’m guessing this bottle neck is because you’re trying to chart far, far too much data. Add LIMIT 1000 to your query and it’ll fly.

3

@k4s1m Thank you so much.

2
2557×744 12.7 KB
1
11462×791 64 KB

It looks like the bottleneck is the data download from the server.
Any suggestion on how to speed that up?

4

That’s progress. Next step:

Do the queries take a long time to run when you execute them on the query screen?

How many rows of data are returned by each?

What happens if you add LIMIT 1000 to the end of the query? Does that improve execution speed?

5

In the query screen, the query takes a few seconds to run, but the visualization takes a long time to load.
The execution speed increases drastically after limiting the row count to 1000.
As a result, I rewrote the query so it will only return about 1400 rows, and I am having much better performance now.