On Tue, 2003-11-04 at 15:34, Mladen Gogala wrote:
You can test it by setting up a table
with PCTFREE+PCTUSED=100. In other word, the answer to your question is that
two parameters are needed to reduce the overhead of the free list maintenance.
Well, geez Mladen, I was trying to make her
I can answer that, or at least try.
When you buy a Mac with OSX, you're getting a system that
is fairly complete, and is supported by the HW vendor.
In contrast, when you install RedHat, ( or any other linux )
for personal use, it can be a real pain in some respects.
For instance: I use
Hi, Gopa:
I ever saw you can rename tablespace via transportable tablespace via edit the dmp
file and reimport metadata into database. But it is not one command line. How do you
rename tablespace in 9i?
I tested rename tablespace in 10i beta1 and found rename tablespace does update
Would you like a multiple choice answer or
is it not obvious? ;)
where was that copy of free QNX again?...
Cheers
Nuno Souto
[EMAIL PROTECTED]
- Original Message -
so now that Linux has made it are we going to have a Linux provider melt
down? with no open source/free versions left?
Akshally, it's a pity Oracle stopped support for Apple's Unix.
My 7.0 version for Apple's A/UX was one of the most stable and
solid ports I ever played with.
If anything, OSX is even more solid and better performing than
A/UX ever was.
Oh yes, the Finder desktop running inside Unix was
Hello all,
I am calling SQL*Plus from a unix shell script and storing the
results of the query executed in a shell variable. It goes like
this:
FL_SUFFIX=`sqlplus -s /nolog EndOfSQL
SET ECHO OFF
SET FEEDBACK OFF
SET VERIFY OFF
SET PAGESIZE 0
Hi,
We are evaluating using either Oracle logical standby or Quest Shareplex
replication for reporting purposes. It appears that there are quite a few
datatypes not supported by Logical standby. Anyone knows what datatypes are
not supported by shareplex replication? Tried looking up at quest
I forgot to mention we're looking at Shareplex 4.5. I found DDLs and
datatypes not supported on version 4.0 but not on 4.5.
Thanks.
elain
From: elain he [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Subject: shareplex: datatype unsupported
Date: Wed, 05 Nov 2003 07:01:44 -0500
Hi,
We are evaluating
On Tuesday 04 November 2003 18:29, Stahlke, Mark wrote:
I'll second that recommendation for Gentoo. I'm running Oracle 9.2.0.3 on
my
Gentoo based laptop and I just installed OWB on it too.
Installing Gentoo can take a long time but once it's done you'll have a
sweet system.
Hello List,
i
if :select_sen_emp_chk_first = 'Y' then
select emp_id
into some_variable ...
from emp
where dept = :dept
and salary :min_sal;
end if;
This might be better
Raj
Rajendra dot Jamadagni at
You have multiple solutions; one would be to use an externally identified Oracle
account; another one to use the hide.c program (should find it with google) to prevent
people from seeing command arguments through 'ps'; another one to write something such
as
myvar=`echo your stuff here
Please bear in mind that there is one thing in a datatype being supported and another
in all functions and features of Shareplex being usable when that datatype is involved.
I am thinking about datatype long specifically.
We have been replicating a 8i database (tru64) to a 9i one (sun)using
Tuesday, November 4, 2003, 9:44:25 PM, Mladen Gogala ([EMAIL PROTECTED]) wrote:
MG Jonathan, you're a very smart guy and a very nice one as well but I cannot
MG make sense of this clarification of yours.
Hey, it's possible that my comments make no sense. I never
promise to make sensegrin.
MG
We are working on a SharePlex project here although we are still in
test/pilot phases. We are in an apps 11i environment with an 8.1.7.4
database. So far, we have found that index organized tables are not
supported. In addition, in financials, there is a table named
hz_locations that has a UDT
The recent article that mentioned sequences got me to
thinking. I might pitch a more detailed article on sequences
to Builder.com. But a more interesting article might be one
that explored various ways to automatically generate primary
keys. So, in the name of research, let me throw out the
I take the easy approach. I use Oracle Designer and when I transform an entity to get
a table, I check the checkbox that indicates surrogate key. It automatically gives
me a sequence generated surrogate primary key. The name of the sequence by default is
the table alias followed by _SEQ so
At one site I worked at, the programmers insisted on using Java
milliseconds as the primary key -- so that they wouldn't have to hit
the database twice (once to get the sequence number, once to insert the
row). They swore up, down and six ways from Sunday that there could
never, ever, EVER be a
Jonathan,
I think your idea of a paper is a good one. But I think we need to back th
question up to what the requirements are.
First, to me, a primary key should not be something that a user would ever
see or use. So the Soc. Sec. # is out. (A side issue - I used to work at a
college. Want to
don't you hate this arguement? Of course, your solution solves the problem
- use triggers to populate the column. it shows that the developers just
don't understand all of the tools that they have available to them.
Tom Mercadante
Oracle Certified Professional
-Original Message-
Sent:
As it was recently discussed,
Insert /*+ append */ into destination_table select * from
source_table
will produce minimum redo/undo if destination_table specified as
nologging.
But, what if destination_table is index-organized table?
Is it possible to achieve the same results (in regards to
Tom,
I think using a natural key such as Soc. Sec. # as the primary key is a good
idea. You don't need to maintain the sequence so there's no performance issue
associated with sequences. There's no issue of gaps. No index root block
contention. It doesn't seem to be industry common practice
Jared Still scribbled on the wall in glitter crayon:
Linux on the desktop is still not for the average user, it's
for geeks only IMO, or those fortunate enough to have a
resident geek to maintain it. ( my family for instance )
and this is the reason my home laptop runs XP pro. i don't have
Social security numbers are notoriously bad natural primary keys. Did you know that
they are re-used? Yes, it's true. Generally, they don't get re-issued until after
one of the users dies, but it's been a problem in the past and still is. What do you
do with people who don't have SSNs?
Mercadante, Thomas F scribbled on the wall in glitter crayon:
Jonathan,
I think your idea of a paper is a good one. But I think we need to
back th question up to what the requirements are.
First, to me, a primary key should not be something that a user would
ever see or use. So the
Rachel,
That's a good case to remember. Java programmers (or architects) sometimes miss
those little things.
I would ask why you used triggers to populate the PK field instead of saying
INSERT ... MYSEQUENCE.NEXT_VAL in the code, or even INSERT ... SELECT ROWNUM
(or ROWNUM+somefixedvalue).
Thanks Stephane,
Your solution 3 has done the job for me.
Regards,
Charu.
-Original Message-
Stephane Faroult
Sent: 05 November 2003 18:10
To: Multiple recipients of list ORACLE-L
'Connected.' message?
You have multiple solutions; one would be to use an externally identified Oracle
ORDER isn't strictly necessary when all you want are unique numbers.
There are times when you cannot afford to lose CACHed values, as John
Kanagaraj has pointed out
in Oracle Applications when generating Cheque numbers. Such sequences
required a patch in
Oracle Apps 10.7 and 11 and/or creation
The only other method that I've seen that hasn't been mentioned is to
generate the primary key of a new row as max(primary_key)+1. Inefficient
as all get out but I've seen it done on small tables with very low
volatility where the business rules required absolute sequentiality. It
worked but
i dont think social security number is actually unique. I heard that there are some
repeats and there are problems with people who are 80 years old drawing money out of
accoutns of people who are 25 due to this problem.
i know its a standard to use SSN as a key, but it might not be accurate.
Yong,
I hope you read the other replies. Soc. is the *worst* use of a PK if there
ever was one.
You say it is a minor problem so it can be easily changed. What if the SOC
is used as the PK/FK in a hundred tables in your system? Is this an easy
change?
The first rule of thumb about PK's is -
My comments [probably off-the-cuff without spending much time
thinking the issues through .?]
1. Hit a table that keeps a counter.
Used to be a mechanism in the Oracle5 days [If I remember correctly,
Sequences came in Oracle6]. Issues were with locking the single
record used as the generator
Actually, I am sorry to hear that. I hoped that you started moonlighting
as an Apple salesman and that every buyer of Cary's book will get a free G5.
It would be nice, don't you think?
On 11/05/2003 08:04:24 AM, Jonathan Gennick wrote:
Tuesday, November 4, 2003, 9:44:25 PM, Mladen Gogala
I think using a natural key such as Soc. Sec. # as the primary key is
a
good
idea.
it is VERY VERY BAD idea.
Oh yea. This is the first thought that these numbers are unique. But the
real life is completely different. Especially where data about people
are involved. So...
We had such a problem
I hate it when Metalink is slow. Clicking the Open TAR
button should not result in a five minute wait.
But I'm just venting whilst Metalink does what it's doing.
Ignore me...
Best regards,
Jonathan Gennick --- Brighten the corner where you are
http://Gennick.com * 906.387.1698 * mailto:[EMAIL
Title: RE: Metalink
Oh... but I like it when it lets you get the just where you click the open tar button... after EVERYTHING has been entered, then can't find the page... THEN can't find your userid and password.
Me thinks that Metalink has developed some of the undocumented features.
It was a compromise... since they had already written their code, I put
in the triggers so that it was transparent to them that the key they
were generating was not being used.
I had to give them something, since I was really trying hard NOT to say
I told you so!
--- Yong Huang [EMAIL
Your right, take a visit at www.lindows.com.
Dick Goulet
Senior Oracle DBA
Oracle Certified 8i DBA
-Original Message-
Sent: Tuesday, November 04, 2003 7:19 PM
To: Multiple recipients of list ORACLE-L
Redhat recommending windoze for desktop.
Hi, Igor,
Direct-path insert does not work for IOTs. This is documented in SQL Reference
for INSERT.
Whether it works for a table without NOLOGGING set (i.e. LOGGING) is not clear
to me. Documentation says the table has to be NOLOGGING, or its tablespace has
to be so. But Tom Kyte seems to show
This isn't a direct answer to your question, but make sure you test
logical standby thoroughly--I had to abandon the idea of using it due
to serious bugs in the apply process, and due to seriously poor
performance of the apply process.
--- elain he [EMAIL PROTECTED] wrote:
Hi,
We are
Hi All,
The network engineer at the site confirmed that all the ports were open.
He played around with it and came up with the following solution.
type in [EMAIL PROTECTED] instead of just username.
I had no problems connecting then.
Hope this helps,
N.
:--Original Message-
Title: Message
Anyone have any experience in synchronizing
Apps HR (11i) information with OID (902)? We are looking at ways to
automatically generate portal accounts out of the HR module. There is documentation
on this in TFM's and it doesn't look all that hard. However,
I talked to
Title: Message
I am writing a procedure to generate a
company calendar and to allow for scheduling various resources in the
organization. This will run as a portlet and I need to verify that the
user has been authenticated through portal before I allow them to
add/edit/delete entries.
Of course, another reason to use a trigger is so that PKs are correctly
generated _regardless_ of the application that's doing the inserting.
--- Rachel Carmichael [EMAIL PROTECTED] wrote:
It was a compromise... since they had already written their code, I
put
in the triggers so that it was
Title: RE: How do you genrate primary keys?
SSAN are not reused by the government at least on purpose. Check it out below:
http://tinylink.com/?WCzYP7kRi2
However there are many other problems with SSANs.
- Sometimes they are accidentally duplicated. Stuff happens when you issue 6 million
That goes for Shareplex too (sorry to state the obvious). I've been
seriously bitten in recent weeks by problems with their stuff too.
_
Tim Onions
Head of Oracle Development
Speech Machines (A MedQuist Company)
...the speech-to-data Application Service Provider
I've got the follwing piece of code which takes almost 3 seconds to run when I execute it on the server itself using sqlplus.
DECLARE p_xml_in LONG(32760); p_xml_out LONG(32760);BEGIN p_xml_in :=
Yong,
M.b. my question was not clear.
I know, nologging doesn't work with IOTs.
What I'd like to know, if there are any tricks (similar to
direct-path) to minimize undo/redo when inserting into IOT.
Igor Neyman, OCP DBA
[EMAIL PROTECTED]
-Original Message-
Yong Huang
Sent: Wednesday,
This comes from $ORACLE_HOME/rdbms/admin/utldtree.sql
I expected to see a dependency here, but did not. Any idea? Im testing because I need
to use these views to write some code.
CREATE OR REPLACE PACKAGE X IS
PROCEDURE Y;
END;
CREATE OR REPLACE PACKAGE BY X IS
PROCEDURE Y IS
BEGIN
Rather than try to get output using the ` characters, see what you can do
with this method:
{
sqlplus -s -XXX
$USER/[EMAIL PROTECTED]
set heading off feedback off trims on lines 300 pages
set whatever else
do this;
do that;
do the other thing;
It's so-called gridlock computing.
On 11/05/2003 09:44:27 AM, April Wells wrote:
Oh... but I like it when it lets you get the just where you click the open
tar button... after EVERYTHING has been entered, then can't find the page...
THEN can't find your userid and password.
Me thinks that
The default SIZE of the histograms is different for ANALYZE and DBMS_STATS.
Be sure you got what you wanted.
Also, as Wolfgang Breitling discusses in his papers, histograms don't deal
with all kinds of skew. For example, if two fields in a table are dependent
(they both show similar/identical
Hit a table that keeps a counter will not scale (will not perform at
high concurrency). It will cause you no end of buffer busy waits
waits, latch free waits for a cache buffers chains latch (even if
db_block_buffers, _db_block_hash_buckets, and _db_block_hash_latches
could be set to infinity),
Yep, in the USA, SSN is very not unique.
Cary Millsap
Hotsos Enterprises, Ltd.
http://www.hotsos.com
Upcoming events:
- Performance Diagnosis 101: 11/19 Sydney
- SQL Optimization 101: 12/8-12 Dallas
- Hotsos Symposium 2004: March 7-10 Dallas
- Visit www.hotsos.com for schedule details...
I have a 9iAS Application Server configuration release 9.0.2 with patch set
9.0.2.1. There is one application server in addition to the
infrastructure. Both reside on the same server. The Discoverer reports
has a one off patch version 53. The database is release 9i version 9.0.1.3
The
Title: External Tables question
I am trying to use external tables, but can't seem to find one thing that I'd like (I have already RTFM'd but may have missed some part).
Is there a way I could load the line number of the text file as a column in the table? line number isn't hard coded but
Cary,
If hitting a table that keeps a counter causes so many performance problems, I
wonder why hitting sys.seq$ is much faster. I'd like to have some education on
this Oracle magic. The only thing I can think of is that Oracle keeps some
numbers in library cache as seen in sys.v$_sequences. Your
Title: External Tables question
What about
rownum?
Waleed
-Original Message-From: Jamadagni, Rajendra
[mailto:[EMAIL PROTECTED]Sent: Wednesday, November 05,
2003 11:25 AMTo: Multiple recipients of list
ORACLE-LSubject: External Tables question
I am trying to use external
You can't load line number into the table except in the case where line number is
actually contained in the table as a column. External tables are great for loading
things into the proper realtional tables but cannot be used for much otherwise.
You cannot index an external table. The best thing
However, since it is a join predicate, the histogram data can not be used.
The CBO uses the density values of the join column(s) to derive the join
selectivity. The density value of a column changes (from 1/num_distinct)
when you collect a histogram. If you create a frequency histogram (aka
im not concerned about the type of join. Im strictly concerned about the join order.
does oracle use histograms and distinctness in determining join order? The odd thing
is that it chose a different join order on these tables earlier and on 'similiar'
joins(ie large number of records and only 4
are histograms only used to determine whether to use an index or join type, not join
order?
From: Wolfgang Breitling [EMAIL PROTECTED]
Date: 2003/11/05 Wed AM 11:49:26 EST
To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED]
Subject: Re: Re: explain plan conundrum
However, since
Hi List,
Does someone throw ligts on the following index behavior
Note
a)name is an unique index column
b) table and index has been analyzed b4 running the query
1) select id from table1 where name like 'ABC%';
FULL Table scan
1) select id from table1 where name like 'AB%';
Index scan
name is
do people actually use a table as a counter these days? Now Im 'assuming' they are
jsut people who dont know about sequences or are there actually 'professionals' who
know about sequencse and decide not to use them.
id assume those tables were used in oracle 5 days because either sequences
cant you use rownum with a 'merge'?
From: Khedr, Waleed [EMAIL PROTECTED]
Date: 2003/11/05 Wed AM 11:34:33 EST
To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED]
Subject: RE: External Tables question
What about rownum?
Waleed
-Original Message-
Sent: Wednesday,
i think thats how mysql does it.
joe
Thomas Day wrote:
The only other method that I've seen that hasn't been mentioned is to
generate the primary key of a new row as max(primary_key)+1. Inefficient
as all get out but I've seen it done on small tables with very low
volatility where the business
I see. Sorry for misreading.
How about direct path load? sqlldr direct=true. But this means your data source
is on the filesystem.
What is M.b.?
Yong Huang
--- Igor Neyman [EMAIL PROTECTED] wrote:
Yong,
M.b. my question was not clear.
I know, nologging doesn't work with IOTs.
What I'd
Well, you have 10053, lev 8 guesses to make.
On 11/05/2003 12:04:26 PM, [EMAIL PROTECTED] wrote:
im not concerned about the type of join. Im strictly concerned about the join order.
does oracle use histograms and distinctness in determining join order? The odd thing
is that it chose a
The join order of an access plan - in the absence of any leading or ordered
hints - is determined strictly like everything else by the CBO: the join
order with the lowest estimated cost wins. And the selectivity and
cardinality estimates play a big role in determining the cardinality and
thus
Having worked for the government in a situation where we were actually
tracking information BY Social Security Number, let me tell you the problems
with it.
1) Social Security Numbers ARE NOT GUARANTEED TO BE UNIQUE
2) Individuals ARE NOT GUARANTEED TO HAVE ONE AND ONLY ONE Social Security
That's it. If you didn't use the cache, then it would cause the same
problem as with normal table-managed sequence numbers. But with cached
sequence numbers, an application can get a sequence number without
touching the database (SEQ$) at all.
Cary Millsap
Hotsos Enterprises, Ltd.
What Oracle version?
Can you post more detail about the table and index.
At 10:09 AM 11/5/2003, you wrote:
Hi List,
Does someone throw ligts on the following index behavior
Note
a)name is an unique index column
b) table and index has been analyzed b4 running the query
1) select id from table1
I go to the 9iAS application server docs page and there are tons of docs. What do i
read first? I flipped through the 'concepts' document and its all over the place.
I dont know what I want to learn, since this isnt for work. I just want to get a feel
for it.
also, I believe i read somewhere
Yong,
If the database is in ARCHIVELOG mode, then the table must be set to NOLOGGING
for append hint to work. If the database is in NOARCHIVELOG mode, then the table
setting does not matter.
Tom has not specified whether the database he tested against was in
NOARCHIVELOG mode or whether the
and it's only slightly better if you have more than one row in that
table. As in, the app the developers here use to generate code keeps a
table of tablenames and their associated last number used
why they felt the need to reinvent the wheel I don't know.
For this app, I couldn't use natural
Yong, sorry but they are federal law prohibiting using SSN as a key, so
the point is moot.
joe
Yong Huang wrote:
Tom,
I think using a natural key such as Soc. Sec. # as the primary key is a good
idea. You don't need to maintain the sequence so there's no performance issue
associated with
Ryan,
hypothetically, When you have a requirement that no gaps allowed in a sequence no
matter what, would you still use sequences?
Raj
Rajendra dot Jamadagni at nospamespn dot com
All Views expressed in this email
Occasionally I see this. It's always a mistake.
I probably see a higher percentage of people that have this problem than
most, because, by the design of my job, practically the *only* systems I
see are ones that have performance problems. Using a table as a counter
is almost guaranteed to cause
Hello Ron
I had a meeting today with people that represent ECORA in Israel.
They have a product called Ecora® Enterprise Auditor
(http://www.ecora.com/ecora/products/enterprise_auditor.asp)
that catalog all your servers and databases.
It can run on your schedule and catalog and produce inventory
There are also rare cases where primary keys are mandated to be consecutive
numbers such that a select of nextval, if it were not used, would invalidate
the key.
Rare, but out there.
Bambi.
-Original Message-
Sent: Wednesday, November 05, 2003 11:09 AM
To: Multiple recipients of list
Unfortunately my source is another table.
By the way (btw.), will sqlldr direct=true work with IOT?
m.b - may be.
Igor Neyman, OCP DBA
[EMAIL PROTECTED]
-Original Message-
Yong Huang
Sent: Wednesday, November 05, 2003 12:25 PM
To: Multiple recipients of list ORACLE-L
I see. Sorry
Thanks, Denny. That's it. I imagine Tom's test database is running in
noarchivelog mode and the tablespace is logging.
Yong Huang
--- Denny Koovakattu [EMAIL PROTECTED] wrote:
Yong,
If the database is in ARCHIVELOG mode, then the table must be set to
NOLOGGING
for append hint to work. If
I'm fully convinced. SSN should not be used as a PK.
Can we also conclude that natural keys in general are only good if you sit in
an ivory tower and do unrealistic lab test?
Yong Huang
--- Bellow, Bambi [EMAIL PROTECTED] wrote:
Having worked for the government in a situation where we were
Sami,
Your problem is not with the index, but rather the cost based optimizer. Most
of us have been beat severely over the head and shoulders through the years that full
table scans are a BAD thing, me included BTW. Well, it's time for the old dog to
learn new tricks. So that I'm
oracle verson 8.1.7
table1 info
=
id varchar2(80) primary key,
name varchar2(50) (unique index on this column)
c1 number
c2 number
c3 number
c4 number
)
-Original Message-
To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED]
Date: Wed, 05 Nov 2003 09:34:25 -0800
What
Histograms are only used to refine the selectivity of a predicate. This in
turn determines the cardinality estimate and various costs such as index
access cost and then of course join costs (NL, sort-merge, and hash) and
join cardinality. This ultimately will drive the decision whether a
Yes, direct-path load works on IOTs, at least in 9.2 running in Solaris 2.8.
Yong Huang
--- Igor Neyman [EMAIL PROTECTED] wrote:
Unfortunately my source is another table.
By the way (btw.), will sqlldr direct=true work with IOT?
m.b - may be.
Igor Neyman, OCP DBA
[EMAIL PROTECTED]
There's six very good reasons listed below to NOT use SSN as your unique
PK, and honestly I can't believe this is STILL an issue for any dba
who deals w/ SSNs. These arguments are YEARS old. Isn't this Data
Modelling 101? I know for sure this exact case is in every text i've read.
How to
No, you cannot. Most entities have natural primary keys. People are the exception
not the rule. I am not advocating the use of natural keys as the primary keys of
tables. I like to sue sequnece numbers for that purpose. However the natural key
should be identified and enforced via a
A trick, use a regular table and create an index that has all the needed
columns.
Waleed
-Original Message-
Sent: Wednesday, November 05, 2003 10:29 AM
To: Multiple recipients of list ORACLE-L
Yong,
M.b. my question was not clear.
I know, nologging doesn't work with IOTs.
What I'd
select *
from
(Select rownum m_id, table_name
from dba_tables)
-Original Message-
Sent: Wednesday, November 05, 2003 12:10 PM
To: Multiple recipients of list ORACLE-L
cant you use rownum with a 'merge'?
From: Khedr, Waleed [EMAIL PROTECTED]
Date: 2003/11/05 Wed AM 11:34:33 EST
You're much too nice.
Rachel
Well, that's not a trick -:)
I wouldn't be asking, if I had enough space for both table and index.
It's a huge narrow table, which never gets updated (only
inserts/deletes) - perfectly fits IOT.
Igor Neyman, OCP DBA
[EMAIL PROTECTED]
-Original Message-
Khedr, Waleed
Sent: Wednesday,
For entity uniqueness you have a unique identifier. You might even have more than
one. For drawing entity relationship diagrams however, I don't know of any tool that
allows you to display more than one, so you have a primary unique identifier and
perhaps other unique identifiers that exist
works and generates redo
-Original Message-
Sent: Wednesday, November 05, 2003 2:04 PM
To: Multiple recipients of list ORACLE-L
Yes, direct-path load works on IOTs, at least in 9.2 running in Solaris 2.8.
Yong Huang
--- Igor Neyman [EMAIL PROTECTED] wrote:
Unfortunately my source is
Thanks Raj and Naveen for your input. However my SQL has a union clause and I want it
to be executed whether select_sen_emp_chk_first is Y/N. I tried the ORDER_PREDICATES
hint suggested by Yong but do not know how to get it to work. Basically from the
explain plan how can we tell when the
I save the beatings for when they truly deserved it. Besides, it was
enough to have them come to me in remorse telling me I was right. From
then on, they never had a design meeting without me there.
--- Thomas Day [EMAIL PROTECTED] wrote:
You're much too nice.
Jeff's paper (and other relevant ones) can be found on Tim Gorman's site
(www.evdbt.com).
Daniel Fink
Goulet, Dick wrote:
Sami,
Your problem is not with the index, but rather the cost based optimizer.
Most of us have been beat severely over the head and shoulders through the
Title: RE: External Tables question
Raj, here is an example of control file:
load data
into table TABLE_NAME truncate
(piece recnum
...
)
Alex.
-Original Message-
From: Jamadagni, Rajendra [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, November 05, 2003 8:25 AM
To: Multiple
Ryan,
Never used PeopleSoft, huh?
Henry
-Original Message-
[EMAIL PROTECTED]
Sent: Wednesday, November 05, 2003 12:09 PM
To: Multiple recipients of list ORACLE-L
do people actually use a table as a counter these days? Now Im 'assuming'
they are jsut people who dont know about sequences
I've never heard of an Oracle sequence not generating unique id's,
OPS/RAC or not. Gaps, yes. Overlaps, never.
Cary Millsap
Hotsos Enterprises, Ltd.
http://www.hotsos.com
Upcoming events:
- Performance Diagnosis 101: 11/19 Sydney
- SQL Optimization 101: 12/8-12 Dallas
- Hotsos Symposium 2004:
1 - 100 of 140 matches
Mail list logo