On 27/05/2010, at 20.00, Josh Berkus <j...@agliodbs.com> wrote:
Well, maybe I'm confused here, but arranging things so that we NEVER
have to visit the page after initially writing it seems like it's
setting the bar almost impossibly high.
That is the use case, though. What I've encountered so far at 3
client
sites is tables which are largely append-only, with a few selects and
very few updates (< 2%) on recent data. In general, once data gets
flushed out of memory, it goes to disk and never gets recalled, and
certainly not written. Thinks are hunky-dory until we reach
max_freeze_age, at which point the server has to chew through hundreds
of gigabytes of old data just to freeze them, sometimes bringing the
application to a halt in the process.
The data doesn't get in there in " no time" if autovacuum was aware of
inserts too it would incrementally freeze the table as it grows.
It would still cause it to be read in again but not in a big chunck.
Couldn't pages that are totally filled by the same transaction, be
frozen on the initial write?
Jesper - given my limited knowledge about how it works.
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
