At least for the Spanner DB, it's good to have a randomly-distributed primary key since it allows better sharding of the data and avoids "hot shards" when doing

At least for the Spanner DB, it's good to have a randomly-distributed primary key since it allows better sharding of the data and avoids "hot shards" when doing a lot of inserts. UUIDv4 is the typical solution, although a bit-reversed incrementing integer would work too https://cloud.google.com/blog/products/databases/announcing-support-for-auto-generated-keys-in-spanner https://cloud.google.com/blog/products/databases/announcing-...