Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: why five (blocks)?

Re: why five (blocks)?

From: Herman de Boer <h.de.boer_at_itcg.nl>
Date: 26 Aug 2003 00:41:21 -0700
Message-ID: <bde5777e.0308252341.7237fd95@posting.google.com>


"Tanel Poder" <change_to_my_first_name_at_integrid.info> wrote in message news:<3f488e5f$1_1_at_news.estpak.ee>...
> Hi!
>
> I've always wondered why the minimum number of blocks in an extent was
> rounded to nearest higher multiple of 5 oracle blocks in earlier server
> versions?
> Nowadays this isn't an issue anymore, but I'm just wondering where does this
> magic number 5 come from?
>
> Tanel.

According to my memory, this is the behaviour the space management layer got in rdbms 7.*. The idea behind it is that in order to avoid fragmentation in the tablespace, all extent requests were rounded up to a multiple of 5 blocks.

However, if there was a need for extent allocation, let's say 63 blocks in size, the first search was for a gap of exactly 63 blocks, and if that did not exist, a 2nd search for 65 blocks was started. Another exception to the common rule was for tiny extents: if the requested size was less then 5 blocks, it was not rounded up.

The behaviour changed slightly from version to version, patch to patch. Rumour existed that the space allocation code was so complex that noone could understand it....

Kind Regards,

Herman de Boer
IT Consultancy Group bv. Received on Tue Aug 26 2003 - 02:41:21 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US