As we design the schema, the questions we must answer are:
- When do we embed data versus linking (see below)? Our decisions here imply the answer to question #2:
- How many collections do we have, and what are they?
- When do we need atomic operations? These operations can be performed within the scope of a BSON document, but not across documents.
- What indexes will we create to make query and updates fast?
- How will we shard? What is the shard key?
