Re: Need some help with SQL case/when performance

On Thu, Feb 25, 2021 at 1:06 PM <smith5646midrange@xxxxxxxxx> wrote:

Are you saying it is slower to update the same location than to write
somewhere else? That doesn't make since to me.

If the reads and writes can be handled by different hardware
(different hard disks, say) then in principle, most of the reading and
writing can happen at the same time.

So, performing N reads and N writes could potentially take as little
as approximately (1 read + N writes) worth of time.

Updating the data in place wouldn't allow that parallelization.
Further, with spinning disks, there could also be additional latency
between each operation.

I'm not saying that this is causing your time differential between SQL
and RPG. I'm just explaining how writing to a different location could
potentially be much faster than writing to the same location.

John Y.