Re: vacuum, performance, and MVCC, and compression

[PFC <lists(at)peufeu(dot)com>]

> What about increasing the size of an existing index entry?  Can that be
> done easily when a new row is added?

	I'd say it looks pretty much like inserting a new index tuple...
	Say "value" is the indexed column.

	Find first page in the index featuring "value".
1	If there is space on the page,
		add the tuple id to the list of the corresponding index entry (just like  
creating a new index tuple, but uses less space).
	else
		look at next page.
		If next page has an index tuple with the same indexed value,
			goto 1
		else
			insert new page and create an index tuple on it

> I would be worried about the overhead of doing that on compression and
> decompression.

	The compression methods mentioned in the article which was passed on the  
list seemed pretty fast. From IO-limited, the test database became  
CPU-limited (and a lot faster).