Re: most expeditious way to update a large number of rows

From: Brian Peasland <dba_at_nospam.peasland.net>
Date: Mon, 16 Oct 2006 17:30:42 GMT
Message-ID: <J78onC.5In@igsrsparc2.er.usgs.gov>

hpuxrac wrote:
> mylesv_at_gmail.com wrote:

>> Thanks, Daniel.  I have five redo logs, each 10MB in size.  Do you
>> think this will suffice?  I'm not concerned with the standby database.
>> I'll let FAL handle that.

>
> You are running a system that's important enough to support with a
> standby that has only 10 meg log files?
>
> Wow.
>
> Is there some history about why those log files were created so small?
> To me, it doesn't make sense to create log files with anything less
> than say multiples of 100's of meg and potentially much larger.
>
> Do you understand the extra work that oracle performs on each log
> switch?
>
> So with your current system you can only generate 50 meg of redo before
> you are out potentially of available space for more activity?
>
> Ouch. If you must stay with 10 meg log files ( I wouldn't ) then
> seriously think about allocating a whole lot more of then. Maybe a gig
> worth?
>

How can you make any assessment of this individual's online redo log size? Did I miss where the redo generation rate was specified? In one system where a 100MB online redo log size is appropriate, other systems may find 1MB more appropriate. The size of the online redo logs should be predicated by the following:

What is the redo generation rate?
If a Standby database is employed, the type of Standby configuration which dictates how transactions are shipped to the Standby database. How does this play into your online redo log size?
How often will archived redo logs be written to tape or other destination not residing on the database server?
In the absence of Standby configuration, what is the SLA for data loss.

One wants to size their redo logs appropriately so that the redo gets archived and copied to a safe location frequently enough so as to meet the defined SLA for that system. However, one does not want to switch too frequently so as to put unnecessary stress on the system. I have some systems which only generate 50MB of archived redo each day. Why would I want a 100MB online redo log on that system? I have other systems which can generate 1GB of redo per minute. IMO, no one size fits all, which is why I enjoy the fact that Oracle lets me determine an optimal size for the online redo logs on a database-by-database basis. :)

Cheers,
Brian

-- 
===================================================================

Brian Peasland
dba_at_nospam.peasland.net
http://www.peasland.net

Remove the "nospam." from the email address to email me.


"I can give it to you cheap, quick, and good.
Now pick two out of the three" - Unknown

Received on Mon Oct 16 2006 - 12:30:42 CDT