(Real) Time series on persistent server

It seems that the dash datatable model is to load the data upon start of the server (globally accessible by callbacks):

init_df = load_data()

app.layout = ...

def some_callback():

BUT, if the data is being updated over time, what is the appropriate way to handle this via dash?

As an example, I want to create a dashboard that shows monthly data about some processes at my job. I start the dashboard and it’s server in July. But I want to show data from August this month. Now, we can create callbacks to get the new data, but that is a slow process and would need to be called every time that someone went to the site.

What is the dash model for updating this data on a persistent server?