Re: _serial_direct_read effect

agonenil_at_gmail.com wrote:

> Hi ,
> I've a situation that I need to read a big table very fast.
> I'm using FTS, but I don't want to use parallel slave (not to over load
> the system). I need to read this data only once
>
> I've tried to play with _serial_direct_read but I've found any
> differnt.
> Alought I think it should reduce cpu usage and latches (not going on
> cache).
>
> My other option is to use recycle pool of none-standard block size.
>
> My questions are :
>
> 1. Is it good to use _serial_direct_read ?
> 2. does someone has experince with this parameter ?

I've read your inquiry and the responses from Jonathan, Sybrand, agonenil, and David and based on the fact that what you are suggesting is undocumented or otherwise unrecommended I would be hesitant to endorse them.

But what bothers me about your inquiry is that other than "academic interest" there doesn't seem to be a business case. Is something slow? How slow compared to some metric? How many records? What percentage of the records do you actually need to read? 1% or 100%? What version of Oracle? And if 10g what does ASH show?

-- 
Daniel A. Morgan
University of Washington
damorgan_at_x.washington.edu
(replace 'x' with 'u' to respond)