On Wed, Mar 30, 2011 at 10:40 AM, Greg Stark <gsstark@mit.edu> wrote:
> But one way or another the hint bits have to get set sometime. The
> sooner that happens the less clog i/o has to happen in the meantime.
I talked about this with Merlin a bit yesterday. I think that his
thought is that most transactions will access a small enough number of
distinct CLOG pages, and the cache accesses might be fast enough, that
we really wouldn't need to get the hint bits set, or at least that
vacuum/freeze time would be soon enough. I'm not sure if that's
actually true, though - I think the overhead of the cache might be
higher than he's imagining. However, there's a sure-fire way to find
out... code it up and see how it plays.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company