RE: Running time of index rebuild

From: Mark W. Farnham <mwf_at_rsiz.com>
Date: Wed, 5 Jun 2013 13:59:40 -0400
Message-ID: <021901ce6216$743fbda0$5cbf38e0$_at_rsiz.com>

Notice that Hans is (correctly) drawing the difference between the order of operations (which will indeed be on the order of n log n) and the details of the time required to perform the required operations.

I didn't follow his forum link and the reference might already be there, but Richard Foote and Jonathan Lewis have this topic pretty well covered. JL has done some substantial digging into things like the width of merge runs and how that affects the process.

mwf

-----Original Message-----
From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Hans Forbrich
Sent: Wednesday, June 05, 2013 1:33 PM
To: oracle-l_at_freelists.org
Subject: Re: Running time of index rebuild

As usual with Oracle, the only answer is 'It Depends'. A lot of variables to consider:

• number of rows involved
• number of blocks that need to be read
• size (bytes) of each component of the index
• since it involves a sort, size of PGA for the rebuilding session
• speed of TEMP tablespace when PGA needs to spill
• is Parallel used (available?)
• other resources being used that might lead to lock/latch/mutex contention

and so on.

There is a lot of good discussion on the OTN Forum at https://forums.oracle.com/forums/thread.jspa?messageID=4269655 including links that may be helpful.

/Hans

On 05/06/2013 8:56 AM, Ingrid Voigt wrote: > Hi, does anybody know how the running time for index create or rebuilds
> inOracle changes withnumber of lines / table size?

--
http://www.freelists.org/webpage/oracle-l


--
http://www.freelists.org/webpage/oracle-l

Received on Wed Jun 05 2013 - 19:59:40 CEST