Comments on: Reasons to use InnoDB Plugin https://shlomi-noach.github.io/blog/mysql/reasons-to-use-innodb-plugin Blog by Shlomi Noach Mon, 10 Aug 2009 18:19:01 +0000 hourly 1 https://wordpress.org/?v=5.3.3 By: shlomi https://shlomi-noach.github.io/blog/mysql/reasons-to-use-innodb-plugin/comment-page-1#comment-3109 Mon, 10 Aug 2009 18:10:09 +0000 https://shlomi-noach.github.io/blog/?p=951#comment-3109 @Sheeri,

As far as I understand, indexes on text columns / BLOBs will be less fragmented internally, as long text data is stored off page (instead of allocating up to 768 bytes per value). This means more data per page, which, apart from less total number of pages, also leads to less splitting, and hence less fragmentation.

I’m not sure how the off-page texts/BLOBs are fragmented.

Otherwise I have no actual information about this. No doubt Vadim (Percona) can tell us all about it.

Regards

]]> By: Sheeri https://shlomi-noach.github.io/blog/mysql/reasons-to-use-innodb-plugin/comment-page-1#comment-3108 Mon, 10 Aug 2009 17:16:32 +0000 https://shlomi-noach.github.io/blog/?p=951#comment-3108 How does Barracuda handle fragmentation? same as InnoDB?

]]> By: Log Buffer #157: a Carnival of the Vanities for DBAs | Pythian Group Blog https://shlomi-noach.github.io/blog/mysql/reasons-to-use-innodb-plugin/comment-page-1#comment-3051 Fri, 07 Aug 2009 16:53:05 +0000 https://shlomi-noach.github.io/blog/?p=951#comment-3051 […] Noach stepped up with reasons to use InnoDB plug-in. “The plugin is a drop-in replacement for “normal” InnoDB tables; enabling many new […]

]]> By: Mark R https://shlomi-noach.github.io/blog/mysql/reasons-to-use-innodb-plugin/comment-page-1#comment-2963 Mon, 03 Aug 2009 10:11:21 +0000 https://shlomi-noach.github.io/blog/?p=951#comment-2963 Yes, compression is the main one; I think fast index creation (and faster index dropping) might be useful in some cases, for example, if you have data partitioned by day and want to create more indexes on old days’ data for efficient searching, but only after the day is over (and the data aren’t as “hot” any more so you don’t really want to do table scans)

]]>