Question

I am creating a database that tracks replays for a game. Each replay has a different game mode that is either team-based gameplay or individual-based gameplay. Depending on the game mode, I want to record the winning team or the winning individual.

I have the following MySQL table that tracks rounds in the replay and the associated winner:

CREATE TABLE replay_rounds (
    replay_id INT UNSIGNED NOT NULL,
    round SMALLINT UNSIGNED NOT NULL,
    winning_player_id INT UNSIGNED,
    winning_team_id TINYINT UNSIGNED,
    FOREIGN KEY (replay_id) REFERENCES replays(id),
    FOREIGN KEY (replay_id, winning_player_id) REFERENCES replay_players(replay_id, player_id),
    FOREIGN KEY (winning_team_id) REFERENCES teams(id),
    PRIMARY KEY (replay_id, round))
CHARACTER SET=utf8
COLLATE=utf8_general_ci
ENGINE=InnoDB;

Using what I have now, if the game mode is team-based then I will set the winning_team_id for each round and set the winning_player_id to null. Likewise, if the game mode is individual-based, I will set the winning_team_id to null.

Performance-wise and in terms of best practice, is it all right to leave it like this? Is there a compelling reason to split this into a separate table for each game mode (even if there’s only two modes)? How about a hypothetical situation where game modes are constantly being added — would this be best resolved by creating a table for every new game mode?

Answer 1

I would just have one winning_id (drop the individual winning ids for team and player) and an additional attribute for game_type_cd. This will be a lookup attribute that is a foreign key to a new table called GAME_TYPE_COES. Then all you need to do is fill in the winning_id and the game type. This will allow for an infinite amount of game types with only having to add data to the GAME_TYPE_CODEs table and not having to change the data structure.