Title: RE: Tricky SQL Question -- Solved
Jonathan,
Thanks for the tips ... let me see how I incorporate this ... things to do ...
1. write a *clever* routine to look at sys.mon_mods$ with dbms_stats.flush_database_monitoring_info to decide which tables to analyze in the next session.
2
Title: RE: Tricky SQL Question
Thanks Steven,
I believe Tom touched on this in his discussion at recent Hotsos conference. My requirements were slightly different, but the logic is still good for me.
Thanks
Raj
-
Rajendra dot
Jonathan Lewis wrote:
>
> Very cute -
>
> But it doesn't really cope well with
> a few outlying values at the top end
> of the range. Using double the count
> to invert the high/low distribution is
> neat - but only if the distribution is
> fairly smooth to start with.
>
> Regards
>
> Jonathan
Very clever !
Can I make a couple of suggestions:
You've got a very large number of tables
in one group - and the startup time for
the analyze might have a big impact on
this group - so how about adding in (say)
one second to the analyze type in order
to cater for startup.
Also - how about taki
Raj,
I may not be offering information useful in solving your specific stats
problem. If that's the case, Undskyld. However, this information is
certainly useful.
This link to the asktom website contains a method for dividing up large
tables into ranges of rowids so that multiple sessions can eff
Title: RE: Tricky SQL Question
Stephane,
Nice ... very nice script ... it is very close to what I came up with.
Thanks everyone
Raj
-
Rajendra dot Jamadagni at espn dot com
Any views expressed here are strictly personal.
QOTD
Very cute -
But it doesn't really cope well with
a few outlying values at the top end
of the range. Using double the count
to invert the high/low distribution is
neat - but only if the distribution is
fairly smooth to start with.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Coming s
pressed here are strictly personal.
> QOTD: Any clod can have facts, having an opinion is an art !!
>
> -Original Message-
> From: Jonathan Lewis [mailto:[EMAIL PROTECTED]
> Sent: Thursday, March 06, 2003 12:44 PM
> To: Multiple recipients of list ORACLE-L
> Subject: Re
Title: RE: Tricky SQL Question -- Solved
Okay,
I cracked it ... if you are interested, read on ... it is not very optimal, but close to what I want. To me 8 streams is standard, so you'd see 8 as hardcoded. Also I found that
select sum(obj_last_analyze_time)/8 from statistics_info
/
Title: RE: Tricky SQL Question
Thanks Jonathan,
I'd like to assign the tables to a group, but need to do that periodically. Also what I do is load all tables that belong to a group in a pl/sql table (bulk updates/bulk collects). That's why I don't want to do read-from-tabl
Thinking back to university days, I think this
was called the knapsack problem, and at the
time there was no algorithm guaranteed to
give an optimal solution.
If there is no simple non-procedural algorithm -
how about a strategy that simply allows each
slave to take the longest task that has not
Title: Tricky SQL Question
Hi all,
I have a tricky situation ... I have a table
columns are
owner varchar2(),
name varchar2(),
ana_tm number
ana_tm represents how much time it took to perform statistics collection for owner.name value. the number ranges from 0 to about 12000 right
12 matches
Mail list logo
