Skip to main content
Software Engineering

You are not logged in. Your edit will be placed in a queue until it is peer reviewed.

We welcome edits that make the post easier to understand and more valuable for readers. Because community members review edits, please try to make the post substantially better than how you found it, for example, by fixing grammar or adding additional resources and hyperlinks.

Required fields*

8
  • 14
    Blaming the mt.gox flaw on surrogate keys seems rather dubious. The problem was that they included all fields in their compound key, even mutable/malleable fields. Commented Aug 16, 2016 at 8:51
  • 7
    A "social" disadvantage of using auto-increment keys is that sometimes "the business" assumes that there must never be any gaps and demands to know what happened to the missing rows that occur when a failed insert happens (transaction rollback). Commented Aug 17, 2016 at 14:51
  • 5
    Another disadvantage is that if the system grows so large that you have to shard the database, you can no longer use autoincrement to produce a globally unique key. When you get to that point, you may have lots of code relying on that assumption. There are other ways to produce a unique identifier that will keep working if the database is sharded. Commented Aug 17, 2016 at 20:37
  • 1
    @Voo It is not guaranteed that your chosen database supports that. And trying to implement it a higher layer than the database itself means you lose some of the guarantees SQL would give you. Finally any centralized assignment of IDs will increase latency if you have a distributed system. Commented Aug 20, 2016 at 0:11
  • 1
    @Voo Of course regardless of the scale of the system one shouldn't make too many assumptions about the nature of autoincremented IDs. If you have just a single database they are assigned in order, but there is no guarantee they are committed in order. And there can be gap in the sequence because not all transactions are committed. Commented Aug 20, 2016 at 0:22