>>> On Fri, Dec 14, 2007 at 5:59 PM, in message <[EMAIL PROTECTED]>,
Tom Lane <[EMAIL PROTECTED]> wrote:
> "Kevin Grittner" <[EMAIL PROTECTED]> writes:
>> Why double writes per read, plus massive writes at checkpoint?
>
> The double writes aren't surprising: freezing has to be WAL-logged,
> and
"Kevin Grittner" <[EMAIL PROTECTED]> writes:
> Why double writes per read, plus massive writes at checkpoint?
The double writes aren't surprising: freezing has to be WAL-logged,
and the odds are that each page hasn't been touched since the last
checkpoint, so the WAL log will include a complete pa
I'm grooming a new server to replace one that is soon to be retired.
Most of the data was loaded very close together, in terms of
database transaction numbers, and probably 95% of it will never be
updated. To assess the potential impact of a "freeze everything in
the database at once" sort of nigh
