I have been using C# to run operations on a DocumentDB instance and really like it so far. I have a lot of C# code that queries from multiple collections to create new collections from the relationships between the first two collections.
Can I essentially move my logic up to the server in stored procedures? I tried answering this question for myself, but all I could find was documentation on how to acquire the collection associated with the stored procedure. So then I thought, could I call a stored procedure that called another stored procedure, passing in the first collection?
Is there any way I can refer to multiple collections in a stored procedure somehow?
Would it be easier to find an easier way to know what belongs to each "collection" if I store everything in the same collection?
Stored procedures run inside of a single collection (or a single partition in a partitioned collection). A call to a stored procedure can only operate on the data in that collection/partition.
When I see this question asked, I usually wonder if you are thinking of collections as a direct analog to tables from the SQL world or even the use of the word "collection" from the MongoDB world. In DocumentDB it's best to not separate your data up by type but rather to mix data of different types in the same collection and separate along some other scale out boundary like tenant, user, geography, etc. If you do that, as long as your stored procedure doesn't need to cross that tenant, user, or geography boundary, it'll be able to provide you with fully ACID cross-document transactions.