This is related to some emails on performance about count(*) being so
slow from a month ago or so...
The issue was that count(*) on something with a unique index should be
able to hit just the index and nothing else, but it can't do this
because it needs to check every tuple to see if it's been deleted.
Would it be worthwhile to store the MVCC transaction info seperately?
This info appears to be either 20 or 26 bytes per tuple, which could be
substantially smaller than the remainer of a row on a wide table.
More interesting would be storing the info in a different format, such
as a listing of rows that info applies to. For OLAP type tables, that
would be a substantial savings.
--
Jim C. Nasby (aka Decibel!) jim@nasby.net
Member: Triangle Fraternity, Sports Car Club of America
Give your computer some brain candy! www.distributed.net Team #1828
Windows: "Where do you want to go today?"
Linux: "Where do you want to go tomorrow?"
FreeBSD: "Are you guys coming, or what?"
