Design the primary key

Let’s consider the different entities, as suggested in the preceding introduction. In the gaming application, you have the following entities:

  • User

  • Game

  • UserGameMapping

A UserGameMapping is a record that indicates a user joined a game. There is a many-to-many relationship between User and Game.

Having a many-to-many mapping is usually an indication that you want to satisfy two Query patterns, and this gaming application is no exception. You have an access pattern that needs to find all users that have joined a game as well as another pattern to find all games that a user has played.

If your data model has multiple entities with relationships among them, you generally use a composite primary key with both partition key and sort key values. The composite primary key gives you the Query ability on the partition key to satisfy one of the query patterns you need. In the DynamoDB documentation, the partition key is also called HASH and the sort key is called RANGE.

The other two data entities — User and Game — don’t have a natural property for the sort key value because the access patterns on a User or Game are a key-value lookup. Because a sort key value is required, you can provide a filler value for the sort key.

With this in mind, let’s use the following pattern for partition key and sort key values for each entity type.

Entity Partition Key Sort Key

Let’s walk through the preceding table.

For the User entity, the partition key value is USER#<USERNAME>. Notice that a prefix is used to identify the entity and prevent any possible collisions across entity types.

For the sort key value on the User entity, a static prefix of #METADATA# followed by the USERNAME value is used. For the sort key value, it’s important that you have a value that is known, such as the USERNAME. This allows for single-item actions such as GetItem, PutItem, and DeleteItem.

However, you also want a sort key value with different values across different User entities to enable even partitioning if you use this attribute as a partition key for an index. For that reason, you append the USERNAME.

The Game entity has a primary key design that is similar to the User entity’s design. It uses a different prefix (GAME#) and a GAME_ID instead of a USERNAME, but the principles are the same.

Finally, the UserGameMapping uses the same partition key as the Game entity. This allows you to fetch not only the metadata for a Game but also all the users in a Game in a single query. You then use the User entity for the sort key on the UserGameMapping to identify which user has joined a specific game.

In the next step, you create a table with this primary key design.