Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Mailing Lists -> Oracle-L -> Re: 64-Bit Oracle on Windows 2003

Re: 64-Bit Oracle on Windows 2003

From: Anjo Kolk <>
Date: Wed, 16 Aug 2006 18:05:57 +0200
Message-ID: <>


Define performance in your case:

- Do you want more users?
- Do you want more tx per user?
- Do you want faster tx per user?

Other Questions:
- How many Raid drives?
- Why Itanium? HP has better boxes with AMDs :-)

If you want more users, then going with 64bit will do that for you (larger shared pool, buffer cache, and more process memory to run all the windows threads), if you would use the same CPU 32 -> 64, you would see a drop of around 10 percent in performance (every thing else should stay the same). The way to improve performance is to reduce physical I/O, by increasing the Oracle Buffer Cache (so 6GB buffer caches are the minimum :-))

Any way, answer the quesions and we can give you a better advise,


On 8/16/06, Freeman, Donald <> wrote:
> Hi, I need some feedback on upgrading to 64-Bit. We are a state public
> health agency. My project team has proposed that during our regular server
> replacement cycle that our next purchase to replace our OLTP RAC system
> should be HP 64-bit servers, each w/ dual 1.6GHz Itanium processors, 16 GB
> RAM, 146 GB RAID drives, dual power supplies, and Windows Server 2003
> 64-bit.
> We are getting a lot of kickback from our director about performance. All
> we can do is speculate about how this is going to work in our environment.
> We have had Oracle in and they told us we'd get up to a five-fold increase
> in performance in an OLTP system. Right now our 32-bit architecture tops
> out at about 60 users. We should be able to respond to a public health
> emergency during which time it should be able to handle a lot more users.
> We think this would get us there.
> Does anybody have any wonder stories about how their life changed after 64
> bit? Here is the exact quote I am trying to address:
> ". I am concerned that we are going to spend over $100,000 for 64 bit
> servers without understanding what we will get for this. I understand this
> should improve performance, but it is not clear that we have a benchmark in
> place. I am not in favor of buying equipment in hopes that performance will
> improve anecdotically. I would like to see some hard numbers that will
> point to whether or not this investment delivers what is promised?"
> Don Freeman
> Database Administrator 1
> Bureau of Information Technology
> Pennsylvania Department of Health
> (717) 703-5782

Anjo Kolk
Owner and Founder OraPerf Projects
tel:    +31-577-712000
mob: +31-6-55340888

Received on Wed Aug 16 2006 - 11:05:57 CDT

Original text of this message