Deploying a Dash app on Render with large Excel sources & heavy pandas merges — what’s the best pattern?

Hi everyone! I have a Dash app that runs perfectly on my laptop, but I’m struggling to deploy it on Render due to data size and startup processing.

What I’m trying to do (short):

Combine several large Excel files once per day to build summary tables and show them in a Dash dashboard.

Problem:

  • Git rejects my large Excel files, so I can’t bundle the data with the app.
  • As I understand, some of my data frames require other data frames to be created, because I merge them all the time. So when depploying, some of the new data frames havent been created yet, so the app crashes. But this dosent happen on desk.

Any suggestions?

Can you upload the excel files to a persistent disk? Persistent Disks – Render Docs

hi @frank123
What’s the size of the dataset and app project?

Hi @frank123, Your intuition to separate data from your Dash app is correct for a production environment. Using Git isn’t a good option, as it’s designed for versioning code, not for serving large data files. I recommend using Cloudflare R2, a service that combines storage and a CDN without charging for data transfer. If your app does heavy pre-processing, you can implement a Cloudflare Worker that runs daily to generate the necessary dataset, so your app only consumes the final resource. Or, if you prefer to keep the dataset pre-processing, you must ensure the final dataset is available before the app renders.