>> But you need Atomicity and/or Isolation (reading the last value
>> and incrementing it are 2 different actions that need to be
>> somehow grouped), so you still need a transaction. But if you go
>> for a HEAP table which you build on application start, the
>> overhead should be far less.
>
> The atomicity and isolation comes from the application-side lock. In
> effect, that function becomes single threaded, and since that's the ONLY
> access to the table, there needn't be any transactional logic on the DB
> side.
cflock is only instance wide.
Jochem
--
I don't get it
immigrants don't work
and steal our jobs
- Loesje
[Todays Threads] [This Message] [Subscription] [Fast Unsubscribe] [User Settings]
