I’m moving to Atlas after years of working fine with mLab.
I am planning to build a MongoDB/Atlas App with 1000s of users who are able to login and order items. I was thinking that should be easy in Atlas but NO!
Welcome to the community forum. As @steevej notes, the Atlas restriction on Database Users is for accounts that log directly into your MongoDB deployment. Atlas Users are able to monitor and manage your Atlas clusters.
Using your original terminology, those are limits on admins rather than a limit on customers.
Direct access to your deployment should be limited to trusted database users connecting from a limited range of whitelisted IP addresses.
End users of your application (who are logging in and ordering items) should have accounts and permissions managed by your application or API.
The same user model should be followed for any database deployments. An application can scale to millions of users, but those are application users rather than database users.
Unfortunately many terms (like “users” and “customers”) are overloaded and depend on context.
From an application point of view, your end users/customers should not not have direct access to underlying infrastructure like database deployments. Irrespective of whether you are using Atlas, your end users/customers should not translate 1:1 to database users. Your applications validate and manage end user requests to data; the only direct database users should be your applications and your DBAs.
The audience for Atlas documentation is your development and admin team, not your customers. From Atlas’ point of view, your admins and applications are Atlas Users and Database Users. Any reference to “customer” in the Atlas documentation is referring to you, as an Atlas customer.