Helma wrote:
> The old machine was a 1 CPU machine. There was an increase of
> performance between 0 to 2 % (if i remember well) , instead of the
> 800% the CEO expected.
>
> So if you want to try the RAM scenario, do as you please. But please
> measure your performance in response times, and not some hitratio. And
> then have a look at the difference in performance.
>
> H.
:D
And there's a bug in 9.2 (patch 6, iirc, and it was HP specific)
that will send your response time skyhigh, because the optimizer
keeps looping the huge buffer - it simply does not understand
searching memory for 6 hours is *not* faster than giving up
after 1 second, and fetching from disk - although it may
be "cheaper".
FvB
|
