At 07:12 PM 11/12/2002 -0500, Bruce Momjian wrote:
>I was unsure how _definiative_ the discussion was.
I am sure that the statements in the document are *not* totally accurate;
but my belief is that they are better than nothing and a good starting
point for tuning.
For example, most numbers are upper limits: the 'Changed' and 'Removed'
figures do not reflect actual free pages, they just reflect pages on which
one or more tuple was added or removed etc. The page may still be 90% full,
and may never make it into the FSM depending on the setting of MAX_FSM_PAGES.
The choice of thresholds is arbitrary (as noted in the document), but also
reflects a good starting point IMO.
I think it is also important to put tuning notes in the main shipping
manuals; do people think it would be a good idea to add a disclaimer that:
tuning is inherently database-specific, you may need to set thresholds much lower, or may find that these are
toohigh - experiment with your database
?
IMO no tuning document can ever be definitive; it can only provide the user
with tools to understand the problem and manipulate the outcomes.
----------------------------------------------------------------
Philip Warner | __---_____
Albatross Consulting Pty. Ltd. |----/ - \
(A.B.N. 75 008 659 498) | /(@) ______---_
Tel: (+61) 0500 83 82 81 | _________ \
Fax: (+61) 03 5330 3172 | ___________ |
Http://www.rhyme.com.au | / \| | --________--
PGP key available upon request, | /
and from pgp5.ai.mit.edu:11371 |/