A data model for follow relationships should be able to answer two questions on behalf of a user:
Who do I follow?
Who follows me?
In Chapter 3, Organizing Related Data, you learned to design our table structures so that all important data access can be accomplished by querying a single partition. For this reason, we're better off considering the above questions separately, and designing the right table schema for each.
We'll start with the question, "Who do I follow?" We'll want a partition per user, with each partition containing all the other users they follow:
CREATE TABLE "user_outbound_follows" (
"follower_username" text,
"followed_username" text,
PRIMARY KEY ("follower_username", "followed_username")
);Simple enough, but there's something unusual here: there are only two columns in the table, and they're both part of the primary key. As it turns out, this is a perfectly valid way to construct a table schema; non-key columns are optional...