Re: SGA_MAX_SIZE vs. SGA_TARGET

From: <frits.hoogland_at_gmail.com>
Date: Sat, 5 Sep 2015 20:30:10 +0200
Message-Id: <6A6F0537-B167-43BD-879B-C81A19E802B5_at_gmail.com>


Verstuurd vanaf mijn iPad

> Op 5 sep. 2015 om 20:07 heeft Dba DBA <oracledbaquestions_at_gmail.com> het volgende geschreven:
>
> Was really getting worried I was senile and was going to start drooling on myself. I knew I was able to get memory to allocate and deallocate in the past. It was on Solaris.
>
> Does anyone know why this varies by unix/linux flavor? What is solaris doing that Linux does not?
Every Unix like platform oracle is on has a HP implementation. Because the implementation is depended on cpu huge page support and the implementation in the operating system, you will see changes in how oracle is able to use it, because it is depended on them.

On Solaris (only I believe) PGA can be allocated in HP.
>

>> On Sat, Sep 5, 2015 at 1:54 PM, Neil Chandler <neil_chandler_at_hotmail.com> wrote:
>> It depends upon your platform. Most platforms allocate max_size so having a lower sga_target is pointless and a waste of memory. Some platforms do not (Solaris), and only allocate sga_target, with max_size an unused top limit.
>> 
>> Neil.
>> sent from my phone
>> 
>> > On 5 Sep 2015, at 18:51, Dba DBA <oracledbaquestions_at_gmail.com> wrote:
>> >
>> > yeah this is old. I know its on the web. However, the responses I see are not to the question I have.
>> >
>> > What is the point to having two parameters? If SGA_MAX_SIZE reserves memory for oracle as an upper bound, but would I want to be able to raise and lower SGA_TARGET? What do I do with the 'spare memory'. PGA_AGGREGATE_TARGET is separate and not taken from memory reserved with SGA_MAX_SIZE
>> >
>> > db_cache,shared_pool, large_pool,streams, java, etc... all come out of SGA_TARGET. So what is the point to this? I am missing something.
>> >
>> > I have I have 20 GB SGA_MAX_SIZE and a 10 GB SGA_TARGET. What is oracle doing with the other 10 GB?
>> >

>
--
http://www.freelists.org/webpage/oracle-l
Received on Sat Sep 05 2015 - 20:30:10 CEST

Original text of this message