Re: 9.2 indexes bigger

Posted in 2000

Mark D. Stock — Wed, 13 Sep 2000 21:16:38 +0100 — source: Informix-list mailing list archive (1991-1998)

Doug Agnew wrote: > > Well, if they are detached indexes, then the index record for the leaf has > to point to the tblspace page for the data as well as the logical row > number. That accounts for 4 more bytes. The other four elude me. Fragment number. Even if there is only one fragment, the index is treated as a fragmented one. Cheers, -- Mark. +----------------------------------------------------------+-----------+ | Mark D. Stock mailto:mdstock@mydas.freeserve.co.uk |//////// /| | http://www.informix.com http://www.informixhandbook.com |///// / //| | http://www.iiug.org +-----------------------------------+//// / ///| | |This email will self-destruct in |/// / ////| | |10 sec. If you received this email |// / /////| | |in error, sorry about the mess. |/ ////////| +----------------------+-----------------------------------+-----------+

↪ replying to Mark D. Stock

Madison Pruet — Wed, 13 Sep 2000 21:21:11 GMT — source: Informix-list mailing list archive (1991-1998)

Not quite..... The partition number + rowid is used if the index has a different fragmentation strategy than the data. If the same fragmentation strategy is used, then we don't even put in the partition number as each index fragment is matched up with a single data fragment. "Mark D. Stock" wrote: > Doug Agnew wrote: > > > > Well, if they are detached indexes, then the index record for the leaf has > > to point to the tblspace page for the data as well as the logical row > > number. That accounts for 4 more bytes. The other four elude me. > > Fragment number. Even if there is only one fragment, the index is > treated as a fragmented one. > > Cheers, > -- > Mark. > > +----------------------------------------------------------+-----------+ > | Mark D. Stock mailto:mdstock@mydas.freeserve.co.uk |//////// /| > | http://www.informix.com http://www.informixhandbook.com |///// / //| > | http://www.iiug.org +-----------------------------------+//// / ///| > | |This email will self-destruct in |/// / ////| > | |10 sec. If you received this email |// / /////| > | |in error, sorry about the mess. |/ ////////| > +----------------------+-----------------------------------+-----------+

↪ replying to Madison Pruet

Doug McAllister — Thu, 14 Sep 2000 09:26:40 -0400 — source: Informix-list mailing list archive (1991-1998)

So the answer is that there are an additional 8 bytes only if the index is a detached one and if it is not detached then it is the same as always? Madison Pruet wrote: > > Not quite..... > > The partition number + rowid is used if the index has a different fragmentation > strategy than the data. If the same fragmentation strategy is used, then we > don't even put in the partition number as each index fragment is matched up with > a single data fragment. > > "Mark D. Stock" wrote: > > > Doug Agnew wrote: > > > > > > Well, if they are detached indexes, then the index record for the leaf has > > > to point to the tblspace page for the data as well as the logical row > > > number. That accounts for 4 more bytes. The other four elude me. > > > > Fragment number. Even if there is only one fragment, the index is > > treated as a fragmented one. > > > > Cheers, > > -- > > Mark. > > > > +----------------------------------------------------------+-----------+ > > | Mark D. Stock mailto:mdstock@mydas.freeserve.co.uk |//////// /| > > | http://www.informix.com http://www.informixhandbook.com |///// / //| > > | http://www.iiug.org +-----------------------------------+//// / ///| > > | |This email will self-destruct in |/// / ////| > > | |10 sec. If you received this email |// / /////| > > | |in error, sorry about the mess. |/ ////////| > > +----------------------+-----------------------------------+-----------+