Re: Detecting xxx fragmentation/corruption?
Date: Fri, 9 May 2008 17:57:15 -0700
Message-ID: <bf46380805091757u4e2400a9ic210cd43d5cb3e77@mail.gmail.com>
On Fri, May 9, 2008 at 5:34 PM, Peter Teoh <htmldeveloper_at_gmail.com> wrote:
>
> Specifically on this - uniform extent size - it will not prevent or
> ELIMINATE fragmentation, right? My understanding is that it will
> only reduced the probability of getting fragmented. The simple
> reason (*Please correct me!!!*) is because when all the extent sizes
> are uniformly the same, it is much easier to allocate/deallocate
> stuff, and thus implement the defragmentation algorithm to reorganize
> it in contiguous format, correct?
>
It eliminates tablespace fragmentation.
>
> > Table/Index fragmentation:
> > When the extents of a table or index are scattered all over a tablespace,
> it
> > is said to be fragmented.
>
> Unlike memory, where access time almost zero, accessing the disk is
> much slower. So sometimes I thought it will be better to spread out
> the data - thus enable simultaneous read by the different heads in the
> disk, just like those RAID design. Ie, fragmentation via
> distribution the blocks out in the disk can improve performance - can
> such things happened?
>
You don't have any control over which track on the disk your data is put. In a carefully crafted test environment you could have this sort of control, but not normally in a production system.
If your data is on RAID, then it is separated across several disks already.
> For FTS scenario, if the blocks are NOT arranged contiguously, BUT
> UNIFORMLY sparsed out somehow, (your fragmentation scenario), the disk
> head may be possible to read ALL THE BLOCKS CONCURRENTLY instead of
> sequentially as in the contiguous design, thus speeding up FTS.
> Possible? (Ie, uniform fragmentation is GOOD)
>
If you had that level of control, there was only 1 session on the database, you might be able to do something like that.
Personally, I think you're overthinking this.
Do you have a specific performance problem targeted?
Or are you trying to determine if you have one?
If the the latter, just ask the users, they will tell you.
-- Jared Still Certifiable Oracle DBA and Part Time Perl Evangelist -- http://www.freelists.org/webpage/oracle-lReceived on Fri May 09 2008 - 19:57:15 CDT