What I’d like to do is store 1 million rows in BT but only pull a small amount of them into a glide table. Then I could do relations and rollups over computed columns.
In my current situation I use a filter in GSheet but because GSheet can’t store millions of rows I have to segment my app for each customer.
No, you can’t.
(I have tested with Big Tables)
The pic shows a Big Query table. Yes, Big Query tables are populated by querying a Google Big Query database.
(I’ve also tested with those).
Time is on my side so if you are saying it’s not possible (yet) I would wait and keep everything under Glides umbrella… else I will move to BQ.
Yeah, I’ve seen your crazy setup 
I think Big Query is probably the direction you need to be taking.
Then you can store all your data in one giant Big Query database, and each customer App can pull in the data for that App using a query.
Although - Big Query does have similar limitations as Big Tables. So that might be a problem. But I do know that Glide are working to remove some of those limitations, so in your case it might be best to wait a while and see how things evolve 
Why bother with BQ or BT… if we have fetch json and javascript columns lol
So… when do we need to use/choose BT vs BQ? 
It’s wayyyy less crazy now thanks to you and others in the community who have taught me best practices. I will take your advice again and wait to see how things evolve.
While I’m waiting I’ll have a go at BQ and educate myself further.
I receive ~1000 rows a day via csv’s in email attachments. The delivery method is not under my control so that’s where I start.
I need to store those rows somewhere
You can store in json, up to 20k rows per cell
I know your point but the problem with using a fetch and json payload is that you need to create a helper table (manually) to transponed the data.
Instead, a native query in BQ is able to create a dynamic table ready to be used.
You create it once… Then just reuse it
Big Tables is a native Glide Data source. It’s just like regular Glide Tables, except… bigger 
It is different in that instead of all rows being loaded immediately to the user device, rows are loaded as they are needed. It also has some limitations around data aggregation, which are documented.
Big Query is what Glide refer to as a “queryable data source”. Data is stored in Google Big Query, and your Glide Tables are defined and populated by queries that you write against that data.
That is my technique for the unlimited rows in glide… But i don’t need bt or bq
You create it once… Then just reuse it
Of course 
but here we need to be efficient, if I create a helper table with 1000 rows and 50 columns (thinking in the worst case) and my payload returns most cases 20-40 rows, we are wasting data and load unnecessary info.
I think I get what you’re selling. I use a technique where I adapted the trebuchet method and the miracle method to save on rows in other areas of my app. Essentially I have a few columns which store csv’s and then I explode those rows by transposing them into a helper table.
Your way does sound more effective
You dont waste anything, my apps are using only 20 rows, to operate on millions, and are faster than Glide
Ok… I have to run… Before they ban me again… 
Thanks!
and how can we create a BT from Glide Editor? There is no such option in the list. 
You need to request access. There is a link at the bottom of this page under FAQ.
@Sean_Martin could you please help our friend @gvalero with Big Table access?