Re: Locking Question (nowait, UPDATE etc)

On Mon, 24 Mar 2003, jeggers_at_tetrix.com wrote:
> I hope this question makes sense:
>
> We have some code (in stored procedures) that selects data FOR UPDATE
> NOWAIT, and it behaves as excepted.

[...]

> For performance reasons, we can NOT do another
> select-for-update-nowait prior to running the actual UPDATE or DELETE,
> this would just be way too slow.

Why can one process be performant while doing the "select for update" yet another process is non-performant while doing the "select for update"?

Is the first "select for update" a single or small update while the second one is larger or bulk update? This seems like some version of the OLTP vs data-warehouse trade-off.

Not really sure if there is a solution other than some amount of application rearchitecture to minimize when the two are clashing.

-- 
Galen deForest Boyer
Sweet dreams and flying machines in pieces on the ground.