Last Tested: Apr 1, 2019
You have two options here.
The first option is to use instant dashboards. With instant dashboards, dashboards display the previous run’s data while new queries are run in the background, letting information appear quickly. This is a labs feature. More information here.
The second option is to persist the explores that the dashboard's tiles are built on. This can be done in the LookML using datagroups with the 'persist_with' parameter or using the 'persist_for' LookML parameter. 'persist_with/for' can be defined for the entire model, or for individual explores.
Note: Instant dashboards need to be turned off for this method to work.
Which to choose?
The benefit to instant dashboards is that we can see old information while new information loads. Instant dashboards will always hit the database, so you can be sure you're looking at fresh data.
The benefit to datagroups is tighter control over your caching given that you know your ETL process. If there is no new data in the database, there's no reason to hit the database again. Datagroups prevent unnecessary queries to the database. Also, some might argue that instant dashboards are misleading because they show stale data at the same time as the new data loads, rather than just showing a loading screen.