On Dec 19, 11:55 am, DA Morgan <damor..._at_psoug.org> wrote:
> Jerome Vitalis wrote:
> > DA Morgan wrote:
>
> >> Jerome Vitalis wrote:
> >>> DA Morgan wrote:
>
> >>>> Yes.
> >>> Did you publish the results anywhere on the web?
>
> >>> TIA
>
> >> Have you read Oracle's license? <g>
>
> > ;-)
>
> > From what we can read on the web about block sharing (e.g.
> >http://www.pythian.com/blogs/282/oracle-rac-cache-fusion-efficiency-a...
> > ), it's difficult to imagine any real-life database whose performance
> > would not suffer (maybe to a limited extent in the best cases) from being
> > turned into a RAC database.
> > The possible advantages provided by the huge global cache are probably not
> > so important in regard of the waits introduced by block sharing. Of course
> > RAC can provide other advantages such as distributed parallel execution.
> > But if a single server already has enough hardware resources (CPU, RAM, IO
> > bandwidth...) to handle the workload, it's difficult to see how a
> > migration to RAC could not hinder the performances (again: with
> > "unpartioned" apps).
>
> > That's why I wanted to know if we could find the results of such test
> > cases anywhere on the web.
>
> Oracle does not sell Oracle as a high performance enhancement. They sell
> it for high availability. Oracle's own published numbers for E-Business
> Suite 11i show scalability at 84%.
>
> What one big-iron box gives you is very high initial expense, very high
> upgrade costs, very high maintenance costs, and the realization that, at
> minimum, you need to purchase two or three of them of which all but one
> will be idle 99.99% of the time to achieve HA.
>
> What RAC gives you is good scalability using inexpensive commodity
> hardware. Incremental and affordable scalability. Highly flexible
> configuration. And most importantly load balancing and transparent
> failover.
>
> If you are going to buy the technology buy it for its intended purpose.
> --
> Daniel A. Morgan
> Oracle Ace Director & Instructor
> University of Washington
> damor..._at_x.washington.edu (replace x with u to respond)
> Puget Sound Oracle Users Groupwww.psoug.org- Hide quoted text -
>
> - Show quoted text -
Most of the databases I have worked on, there are never more than 100
concurrent users. Typicaly one server is more than enough. Second
server used for DR or high availblity is also used as a QA box, so it
is not wastful (or doing nothing 99.9% of the time). While RAC is a
good technology for scalability, it is difficult to find applications
where RAC is better solution than non RAC.
Received on Sat Dec 22 2007 - 22:57:16 CST
