-
Type: New Feature
-
Resolution: Unresolved
-
Priority: Major - P3
-
Affects Version/s: None
-
Component/s: Index Maintenance, Performance
-
None
-
Query Execution
A great way to add some relational-ness to the database and give it more powerful features using data already in the system would be to implement a method to define nodes and vertices of a graph through some definition or configuration of an existing collection or set of collections, much the way you'd define an index.
Currently many of us are using another or our own custom build graph databases to get around quickly solving relational shortfalls well identified in mongo. Maintaining the same data in two systems for this could easily be overcome and improve the lives of many.
In the same way that you've improved your database by adding text indexing, collection / document locking, etc, it would be great to see this layer added on top for those of us with wildly complex relationships to create, but for whom a document database is still substantially better than a traditional relational SQL database.
There are others in the document db space doing this, but your document and api implementation are vastly superior. It would be great to see you give some method for self-defined n-tiered relationships through a graph style while keeping the query language as simple as possible.