The database is mostly an abstraction layer for other people to get up and running with the API. On ZipLocate.us everything is cached in memory using Varnish, so API calls never even hit the database. When data is updated, updated zip codes are just refreshed in the cache.
Like disdev, I always use an auto-incrementing id, but you're right that the index should be on the zip code instead of the id.
Like disdev, I always use an auto-incrementing id, but you're right that the index should be on the zip code instead of the id.