5 ms·

> All of our existing indexes were based on string data which had associated integer IDs in our backend You already have a perfect hash function :).

• greiskul 12y ago
  The IDs might be sparse though.
  • sltkr 12y ago
    Sparse is fine if they are more-or-less randomly distributed.
    • awda 12y ago
      I was mostly aiming for unique -> no hash collisions.