comp.databases.oracle.server Google Group

Re: Why is this query slow?

hooperc2...@yahoo.com (Charles Hooper) — Fri, 12 Mar 2010 13:46:23 UT

On Mar 12, 5:34 am, vsevolod afanassiev
The 10% rule is not suitable in all cases. Last year I posted to this
group a test case that showed a full table scan was faster when
selecting 0.06% of the rows in a 100,000,000 row table than was an
index access when the index had a high clustering factor. However,

Re: Why is this query slow?

orat...@msn.com (ddf) — Fri, 12 Mar 2010 13:46:14 UT

On Mar 12, 5:34 am, vsevolod afanassiev
How does this rewrite perform:
with tlimit as (
select ID
from transaction
where tx_date between (date3 and date4)
and status = 'AAA'
)
select * from transactions
where tx_date between (date1 and date2) and
status = 'AAA' and
ID not in
(select ID from tlimit);

Re: New Procedurs With Version# In Name

zifnab...@gmail.com (scottgamble@comcast.net) — Fri, 12 Mar 2010 13:01:18 UT

In a previous life I was in a situation where something similar had
been done.
There were 2 ids for each application. The ids had private synonyms to
a version of the
stored procs.
When functionality of the procedure was changing a new version went
through all the test cycles etc and was put in with a new version

Re: Multiple selects in one

mah...@web.de (Randolf Geist) — Fri, 12 Mar 2010 12:19:42 UT

I think what you're looking for is commonly called a PIVOT operation
that turns rows into columns. Oracle 11g introduced the PIVOT
operator, but it can be done with pre-11g versions as well by using a
GROUP BY with a bit cumbersome aggregate function expression.
Something like this should give you a starting point:

Re: New Procedurs With Version# In Name

jeremy0...@gmail.com (Jeremy) — Fri, 12 Mar 2010 12:17:54 UT

In article <pan.2010.03.11.16.16...@email .here.invalid>,
n...@email.here.invalid says...
What *is* the recommended method? I would probably do
"select count(pk_id_column)..."

Why is this query slow?

vsevolod.afanass...@gmail.com (vsevolod afanassiev) — Fri, 12 Mar 2010 10:34:02 UT

I have been doing SQL query tuning for a while and in most cases this
is relatively straightforward excercise. For example: if a query
selects only a few rows from large table then it should be accessed
through index, if a query selects more than 10% of the rows the table
should be accessed through full table scan. In many cases I was able

Re: New Procedurs With Version# In Name

what...@xs4all.nl (Shakespeare) — Fri, 12 Mar 2010 09:12:53 UT

Op 11-3-2010 6:17, Mladen Gogala schreef:
And I thought SourceSafe was a versioning system....
Versioning systems won't help if both versions of the procedure must be
kept in the software, indeed because other programs may need the old
version when they can not handle changes made to the original procedure.

Re: Multiple selects in one

what...@xs4all.nl (Shakespeare) — Fri, 12 Mar 2010 08:11:52 UT

Op 11-3-2010 17:49, The Magnet schreef:
Man, I sure hope I never have to do any maintenance on code like this....
Shakespeare

Re: Cannot change ARCHIVELOG-MODE via EM

what...@xs4all.nl (Shakespeare) — Fri, 12 Mar 2010 08:03:03 UT

Op 11-3-2010 12:07, gazzag schreef:
Well, I'm back. I was wrong. Sorry! Strange enough, I do not recall ever
have used the " with Windows. I'm getting old....
Shakespeare

Re: New Procedurs With Version# In Name

galen_bo...@yahoo.com (Galen Boyer) — Fri, 12 Mar 2010 01:58:16 UT

If you do this then you need to treat your codebase just as though you
have multiple versions of an application in production at different
client sites.
You need a different schema for each "version" of your production copies
and each schema needs to own all of the code from the Source Control
which represents this version. You will need private synonyms from your

Re: New Procedurs With Version# In Name

n...@email.here.invalid (Mladen Gogala) — Thu, 11 Mar 2010 18:41:16 UT

There probably is a PHB in his company.

Re: New Procedurs With Version# In Name

joel-ga...@home.com (joel garry) — Thu, 11 Mar 2010 17:17:41 UT

On Mar 10, 9:07 pm, "jeffchi...@gmail.com" <jeffchi...@gmail.com>
wrote:
Even Oracle can't get this right. The first two off the top of my
head, statistics generation and job control.
It does make sense for major functionality changes, though Vladimir's
point about overloading is significant for more modern versions of

Re: Oracle 10g on Solaris 10 non-global zones with asynchronous I/O

joel-ga...@home.com (joel garry) — Thu, 11 Mar 2010 16:49:25 UT

There are a bunch of MOS notes on solaris setall, including one about
a mandatory patch. It varies by Oracle version/patch level, which I
can't seem to find in your post. Start with How To Verify Whether
DIRECTIO is Being Used [ID 555601.1]
jg

Re: Multiple selects in one

a...@unsu.com (The Magnet) — Thu, 11 Mar 2010 16:49:14 UT

Well, finally came up with this beast. It works, so, I'll go with it:
SELECT customer_id, non_ut_subscr_id, non_ut_product_id,
non_ut_status, ut_subscr_id, ut_product_id, ut_status
FROM (SELECT customer_id, subscription_id non_ut_subscr_id, produst_id
non_ut_product_id, status non_ut_status,
LEAD(subscription_id, 1, NULL) OVER (PARTITION BY

Re: New Procedurs With Version# In Name

johnbhur...@sbcglobal.net (John Hurley) — Thu, 11 Mar 2010 16:44:33 UT

On Mar 11, 12:07 am, "jeffchi...@gmail.com" <jeffchi...@gmail.com>
wrote:
snip
Getting overruled by who exactly?
Is there any kind of change control team in your company?