Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Mailing Lists -> Oracle-L -> RE: how to monitor the progress of inserts
Here is what I use:
Eric Yen
-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Wolfgang Breitling
Sent: Tuesday, February 01, 2005 5:09 PM
To: Mladen Gogala
Cc: MGogala_at_allegientsystems.com; 'Niall Litchfield'; Oracle-L (E-mail)
Subject: Re: how to monitor the progress of inserts
At 03:50 PM 2/1/2005, Mladen Gogala wrote:
> > As Niall demonstrated so beautifully, there are no relational rules
when
> > it comes to v$ views (actually the x$ "tables" upon which the views
are
> > built).
>
>Yes, but then this statistics is meaningless. What does it mean to have
>1000 inserts without any consequence to the table?
You have to use the v$ views in the context they are meant for. This=20
particular table is meant to help dbms_stats to decide whether the=20
statistics are "stale" and should be refreshed. It is NOT an audit view.
If=20
you talk about useless views this is a truly useless view, even if it=20
didn't suffer from the demonstrated shortcoming. What does it mean to
have=20
1000 inserts, even if those inserts actually made into the table??=20
Statistics could be "stale" if much less than 10% of the content has=20
"changed" and for other tables nothing changes materially in the
statistics=20
even if you exchange 100% for new data. I would never base statistics=20
gathering on the content of this view.
PS. there are people who are very knowledgeable about Oracle who
consider=20
the data in many of the "cherished" v$ views as useless, meaningless,
even=20
misleading.
>I wouldn't generalize just like that. I don't find V$SESSION_LONGOPS
very=20
>useful,
>period. It has never helped me with anything. It most certainly did not
>help me
>to see how much is left of any large job. While V$ tables in general
are=20
>remarkably
>useful, this particular one is useless.
Obviously our mileages vary. I use it frequently to see the progress of=20
full scans and even calculate the estimated finish time. I generally
find=20
it more precise than Microsoft's progress bar. And as I said. if the=20
progress is not linear but slows down ( who cares if it speeds up :-) )=20
then that is a good piece of information as well to go and find the
cause=20
for the slowdown.
Regards
Wolfgang Breitling
Centrex Consulting Corporation
www.centrexcc.com=20
-- http://www.freelists.org/webpage/oracle-l -- http://www.freelists.org/webpage/oracle-lReceived on Wed Feb 02 2005 - 18:07:26 CST