http://www.rittmanmead.com/2005/01/05/compressed-composites-oracle-10g-compression-explained/ Delivered Intelligence Mon, 15 Mar 2010 14:12:52 -0400 http://wordpress.org/?v=2.8.4 hourly 1 http://www.rittmanmead.com/2005/01/05/compressed-composites-oracle-10g-compression-explained/comment-page-1/#comment-236 Chris Chiappa Fri, 07 Jan 2005 20:51:00 +0000 http://www.rittmanmead.com/wp2/2005/01/05/compressed-composites-oracle-10g-compression-explained/#comment-236 …of course, I’m speaking for myself, comments should not be construed as company positions, these are all opinions, etc, etc, etc. …of course, I’m speaking for myself, comments should not be construed as company positions, these are all opinions, etc, etc, etc.

]]> http://www.rittmanmead.com/2005/01/05/compressed-composites-oracle-10g-compression-explained/comment-page-1/#comment-235 Chris Chiappa Fri, 07 Jan 2005 20:48:00 +0000 http://www.rittmanmead.com/wp2/2005/01/05/compressed-composites-oracle-10g-compression-explained/#comment-235 Mark, the compression benefit does apply only when you tend to have “sticks” - for instance, if JAN, FEB, MAR roll up into Q1, but only FEB exists. We’ve seen that most data these days is sparse enough that compression is usually a benefit, and when it doesn’t help, it usually doesn’t hurt either. And when it does help, it’s often huge - we saw one customer data set (from manufacturing) where it brought build time down from about 2 hours to a scant 15 minutes. The “limited” application is more about restrictions that come into play when you’re using compressed composites - in the initial release you can only aggregate compressed composites using SUM, and you can’t write data into a variable dimensioned by a compressed composite that has been aggregated without first clearing out the aggregated data. There are probably others, but those are the ones that jump to mind first. Like all features, compressed composites are being continually improved and you can be sure that future releases will relax or remove many of the current limiations. Mark, the compression benefit does apply only when you tend to have “sticks” – for instance, if JAN, FEB, MAR roll up into Q1, but only FEB exists. We’ve seen that most data these days is sparse enough that compression is usually a benefit, and when it doesn’t help, it usually doesn’t hurt either. And when it does help, it’s often huge – we saw one customer data set (from manufacturing) where it brought build time down from about 2 hours to a scant 15 minutes.
The “limited” application is more about restrictions that come into play when you’re using compressed composites – in the initial release you can only aggregate compressed composites using SUM, and you can’t write data into a variable dimensioned by a compressed composite that has been aggregated without first clearing out the aggregated data. There are probably others, but those are the ones that jump to mind first. Like all features, compressed composites are being continually improved and you can be sure that future releases will relax or remove many of the current limiations.

]]>