[development] Why do we avoid auto-incrementing db columns?

Boris Mann boris at bryght.com
Mon Apr 30 06:55:08 UTC 2007

On 4/29/07, Sammy Spets <sammys-drupal at synerger.com> wrote:
> Earnest,
> Insertion of 20 million nodes in a short space of time implies 10
> million users are inserting 2 nodes each in a short space of time. It
> also implies many other combinations. Looking at the plausibility that
> this situation become true.... pretty much no chance you'll get that
> many insertions in the same table at the same time except for the
> watchdog table.

Uh. Except for high performance external sources of data feeding a
Drupal front end. Remember what assumptions make out of you and me...

Earnest figured it out because he needed to implement it for a project.

> How does your technique of randomly allocating IDs avoid duplicate ID
> allocation?

I believe it checks for dupes and re-randomizes, reading his write up.

Boris Mann
Office 604-682-2889
Skype borismann

More information about the development mailing list