hpuxrac wrote:
> DA Morgan wrote:
>> hopehope_123 wrote:
>>> Hi ,
>>>
>>> We are about to move our current datawarehouse system from redhat
>>> linux to a sun solaris box. ( 6 dual core cpus, 2 hba , emc disk
>>> array , raid 10 , 24GB. RAM)
>>>
>>> The db size is about 2tb.
>>>
>>> Since this is a datawarehouse system , random sequential io is my main
>>> concern , we also use parallel query heavily.
>>>
>>> I have some questions on my mind:
>>>
>>> 1. What file system to choose? UFS , Veritas or raw device?
>>>
>>> 2. If i choose ufs or veritas , should i use direct_io or not? Since
>>> parallel query bypasses the sga , and directIO bypasses the page
>>> cache , using buffered io is better .
>>>
>>> 3. How can i coose the strip size of the raid?
>>>
>>> 4. In order to maximize the sequential io performance , i am thinking
>>> of setting db_file_multiblock_read_count to 1MB. ( Which provides 1MB.
>>> read on each disk read) , along with the unix kernel parameters
>>> maxcontig (for ufs) etc. How can i balance this number with the strip
>>> size? Is 1MB. maximum?
>>>
>>>
>>> I will be appreciated if you can comment , advise.
>>>
>>>
>>> Kind Regards,
>>> hope
>> I'd recommend RAW managed with ASM.
>>
>> Would recommend you avoid third-party products that will inevitably
>> lead to higher costs, training issues, and vendor finger-pointing
>> exercises.
>
> Are you saying that you did not write the above words?
>
> Then someone must have stolen your google groups login.
>
>> --
>> Daniel A. Morgan
>> University of Washington
>> damorgan_at_x.washington.edu
>> (replace x with u to respond)
>> Puget Sound Oracle Users Group
>> www.psoug.org
I don't use google groups. Ever!
--
Daniel A. Morgan
University of Washington
damorgan_at_x.washington.edu
(replace x with u to respond)
Puget Sound Oracle Users Group
www.psoug.org
Received on Mon Jun 19 2006 - 14:59:27 CDT
