* Don Barthel <dbarthel at usedeverywhere.com> [070415 11:02]:
> Thanks to Bill, Andrew and Jan for their advice!
> 
> Based on the feedback I have decided on a new solution.
> 
> Again, I have a main database and a slon replicated copy that I use
> for reporting. But the table I report from is big and unwieldy and I
> want to speed up the reports as much as possible.
> 
> So I have created a new table that will contain a subset of the rows,
> a subset of the columns and in some cases a subset of an individual
> column when I don't need the whole column (i.e. my reports only need
> the first 200 characters of the description). And I'm maintaining the
> new table with triggers.

Sounds fine, I just wanted to remark, that PostgreSQL deals rather
well with just leaving out the unused columns out of the select.

So I wonder have you tried leaving out the unneeded columns, and
creating partial indexes for the subset of rows you are interested in?

Andreas