Fastest implementation of storage for multi page apps?

Dash Python
1

Hi all,

I’m mostly looking for best practices here.

I’ve got a multi page app for all my company KPIs. In this, I’ve previously had multiple “Global” DataFrames that get filtered on each page. I’ve currently got those stored on my Heroku instance as pickles. Things are ~somewhat fast, but I’d like them to be faster.

I’ve worked and considered:

  • DCC.Store after loading everything on my index page, and then reading JSON from there (this seemed to slow things down)
  • Server Side Storage (but haven’t resolved why it’s not working for me, Show and Tell - Server Side Caching - #96 by Emil)
  • Storing on Postgres DB and calling that on each page, or using the same DCC.Store methodology

Just wondering in general if there were any best practices for a relatively large data set that has to be persisted across ~6 pages.

2
  1. Will be slow, unless you use clientside callbacks
  2. Should be about the same speed as you current solution
  3. Depends on query pattern, can be faster (if you query a subset of the complete dataset), or slower (if you always query it all)

From a performance perspective, I think your current approach is OK. If you need more speed, a Redis cache might be an option.

1 Like