Hi Rachel,
Correct, Skip Scan Index is not a type of index but a method whereby
Oracle can eliminate the need to visit leaf nodes by determining whether the
leading column(s) have changed by sussing out only the branch nodes. It's
possibly useful in situations where previously Oracle would not
I tried it and what it does is, essentially, a fast full index scan on the
remaining columns of the index. To resolve the query, oracle does a full
sequential scan on the index instead on the table. If your index is one third
size of the table, you saved quite a few IOs but don't expect anything
Mark,
thanks, interesting article. But I'm still getting the feeling that the
index skip scan is helpful only when you don't want to create a
secondary index on columns that are not the left-most column. since I
believe that we will be doing a LOT of queries by order date as well,
I'm not sure
If anybody knows, it must be Lex...
Lex, you genius of geniuses and SQL Logician of them all - can you help here?
Mogens
Mark Richard wrote:
This is quite close to what I was thinking of however it doesn't seem to
work for Oracle. Does anyone know if there is similar syntax available in
Title: RE: skip scan index
Rachel,
Skip scan index is not a index type, it is a index scan type. Maybe the developer should re-read the relevant portion of the manual. If your order volume is low, you probably won't see much performance impact by having two indexes (like you need to hear
It's an oxymoron, I'd say.
Gogala, Mladen wrote:
What is a predictable corporate environment? I know of no such
thing.
Mladen Gogala
Oracle DBA
Phone:(203) 459-6855
Email:[EMAIL PROTECTED]
-Original Message-
Sent: Tuesday, May 27, 2003 4:50 PM
To: Multiple recipients of list ORACLE-L
If you check out product availability on MetaLink there is NO 9i 32 bit for HP.
It's all 64 bit.
Dick Goulet
Senior Oracle DBA
Oracle Certified 8i DBA
-Original Message-
Sent: Tuesday, May 27, 2003 9:10 PM
To: Multiple recipients of list ORACLE-L
On Tue, May 27, 2003 at 03:19:40PM
H ... Oracle 11 h beer
http://www.oracle.com/appsnet/technology/upgrade/docs/features.html
JP
On Wednesday 28 of May 2003 12:07, you wrote:
Hi
Can anybody tell me a website where I can know about new features in Oracle
11/11i?
Thanks in Advance
Ajay K. Garg
--
Stephane,
The queries will almost always include data from the data blocks in
addition to the index information. The only query I can think of where
that would not be true would be a simple count of orders by date.
I'm beginning to think, based on this discussion and others we've had
here where
1. If I analyze the code using the explain plan,
where is info created from this explain plan stored
at?
Look for the PLAN_TABLE table (prior to 9 - otherwise there is a package). If you
don't have any PLAN_TABLE create it by running $ORACLE_HOME/rdbms/admin/utlxplan prior
to executing the
Got it this is going to buy me exactly nothing given my app and
database design and database size..
--- Mladen Gogala [EMAIL PROTECTED] wrote:
I tried it and what it does is, essentially, a fast full index scan
on the
remaining columns of the index. To resolve the query, oracle does a
At 02:59 AM 5/28/2003 -0800, you wrote:
Okay, I have a developer here who has been reading the docs (this can
be dangerous!)
we are adding functionality to one of our applications, this will
involve using multiple fulfillment houses, so we'll be adding the
fulfillment vendor id to the order table.
Rachel,
I'll send you a baseball bat to club that duhveloper over the head with. Your
right, skip scan is a method that Oracle uses to make use of an index when logically
it should not. You cannot specify it that way. Darn duhvelopers who read things into
manuals, it's dangerous for
Hi Rachel,
Correct, Skip Scan Index is not a type of index but a method whereby
Oracle can eliminate the need to visit leaf nodes by determining whether the
leading column(s) have changed by sussing out only the branch nodes. It's
possibly useful in situations where previously Oracle would not
I don't doubt that it works I just doubt that I NEED it :)
--- Jamadagni, Rajendra [EMAIL PROTECTED] wrote:
Rachel,
Skip scan index is not a index type, it is a index scan type. Maybe
the
developer should re-read the relevant portion of the manual. If your
order
volume is low, you
Richard,
the access would be order date or vendor id/order date (since it's
possible to look up by vendor id alone as well)
very low cardinality on vendor id -- right now I have all of two.
personal opinion is that the developer read something cool and
decided to tell the DBA how to do things,
Rachel
My experience with index skip scans can be summed up as follows. If you know
the app will be doing a particular scan, create the index.
Index Skip Scans should be thought of a means to help optimize those pesky
ad-hoc queries only.
I haven't been able to get a skip-can to work unless
Hi all,
I know this is RTFM but I can't find the right part in the FM to R. So any
help would be appreciated:
In the context of:
DBMS_STATS.GATHER_DATABASE_STATS(method_opt='FOR ALL HIDDEN COLUMNS');
what is a hidden column. I tried gathering stats FOR ALL HIDDEN COLUMNS
and it didn't
Dick,
I club him regularly... doesn't seem to get through. As someone (Bill
Thater) once said on the OT list.. this guy wouldn't be able to spot a
clue, in a clue field, during clue mating season while drenched in clue
pheronomes.
I fight all the time with him. I win :). My basic premise is
Thanks for everyone for responding. I think I have enough to get as close
as I can to what was requested.
Kevin
-Original Message-
Sent: Tuesday, May 27, 2003 5:40 PM
To: Multiple recipients of list ORACLE-L
I'll beg/borrow steal from John Beresniewicz's book on Oracle Built-in
delete table1
where exists
(
select 1
from (
select column from table_a
union all
select column from table_b
) a
where a.column = table1.column
);
-Original Message-
Sent: Wednesday, May 28,
Sorry for wasting bandwidth - found the answer 5 minutes later (found it by
looking up histograms in the manual, not DBMS_STATS):
For those who want to know - a hidden column is an expression for a
function based index. Therefore you can use this option to gather stats
when using function
I have a question about index creation. When I run
create index xyz ... tablespace indx_ts ...
at what stage sort_area_size (and temp space if sort_area_size is not
large enough) is used during the process? Is my understanding correct that
the index will be created in indx_ts (not temp
We have two servers: Server A running iAS ; Server B running IDS 9 and
serving files.
We map one drive from server B to server A and we are trying to access the
forms on the remote driver:
http://host:port/forms90/f90servlet?form=f:\form_name.fmx
But we receive an error saying that the servlet
Answer to question #1: The plan_table owned by system or sys.
#2: Try the following: V$BUFFER_POOL, V$LIBRARYCACHE, V$DB_OBJECT_CACHE,
V$MYSTAT, V$SESSION_WAIT, V$SESSTAT
Good Hunting.
I'll take your vote! Especially since you have hard evidence that you
can't always get there from here and even with one value (we'll have
two at the beginning) get an index skip scan to occur
--- Wolfgang Breitling [EMAIL PROTECTED] wrote:
At 02:59 AM 5/28/2003 -0800, you wrote:
Okay, I have
Rachel,
For skip scan to work you'll need statistics, namely CBO, which if that's the
case, create the one index if needed due to a unique constraint and forget about it.
Chances are that the CBO will decide on a FTS anyway.
Dick Goulet
Senior Oracle DBA
Oracle Certified 8i DBA
What, like:
Oracle, Unbreakable?
;)
-Original Message-
Nrgaard
Sent: 28 May 2003 14:05
To: Multiple recipients of list ORACLE-L
It's an oxymoron, I'd say.
Gogala, Mladen wrote:
What is a predictable corporate environment? I know of no such
thing.
Mladen Gogala
Oracle DBA
Kevin,
Thanks these will NOT be ad-hoc queries but part of the app -- for
the admin and customer service users.
I'm leaning more and more towards setting things up so that we either
do a full table scan or use two indexes.
I just did a query -- since the app was released in December, we
Rachel,
First, I'd strongly suggest you look at 9.2.0.3... there are a number of bug
fixes in it and we have been running it here for a couple of months now I
would say with few problems.
With regards to the skip scans on indexes, your assumptions are correct.
I've seen some cases where skip
We have two servers: Server A running iAS ; Server B running IDS 9 and
serving files.
We map one drive from server B to server A and we are trying to access the
forms on the remote driver:
http://host:port/forms90/f90servlet?form=f:\form_name.fmx
But we receive an error saying that the servlet
List - If I wanted to know whether my query was taking advantage of index
skip scans, how would I know? Is there something different in the EXPLAIN
PLAN that I should look for? The discussion just made me curious.
Dennis Williams
DBA, 80%OCP, 100% DBA
Lifetouch, Inc.
[EMAIL PROTECTED]
Hello everyone,
I have a question for the group of which method is more efficient.
To set the stage my company has a process to load part changes from vendors
into the tables in an 8.1.7.4 Oracle database with archiving on and this
database has a standby database at disaster recovery site, so
Title: Unbreakable, my buns. Upgrade to 9.2/w2k
For those of you who have upgraded to 9.2 on w2k:
How many of you have ended up with such a hosed server that you had to completely wipe Oracle off of it and start with a fresh install (9.2)? It was spectacular! I ended up in this situation
Bill - Try Google. A quick search yielded the fact this has something to do
with creating columns of objects, nested table, varray, or REF type.
Dennis Williams
DBA, 80%OCP, 100% DBA
Lifetouch, Inc.
[EMAIL PROTECTED]
-Original Message-
Sent: Wednesday, May 28, 2003 9:45 AM
To: Multiple
Title: Message
Hi
all,
I have a database
where all tables and indexes have maxextents set to unlimited. Why, in
dba_tables, do some tables have a null value for maxetents and some have a value
of2147483645?
An index I just
modified (previous value of maxextents was 249) now has a null
Hidden columns are used if you use certain features of Oracle, e.g.
materialized views, replication, objects
At 06:44 AM 5/28/2003 -0800, you wrote:
Hi all,
I know this is RTFM but I can't find the right part in the FM to R. So
any help would be appreciated:
In the context of:
A short cut to test the new feature is using the hint index_ss(table,index).
Index skip scan is not an index scan or fast full scan.
Regards,
Waleed
-Original Message-
Sent: Wednesday, May 28, 2003 7:00 AM
To: Multiple recipients of list ORACLE-L
Okay, I have a developer here who has
Yes, the OPERATION column in PLAN_TABLE will show something along the lines
of index access - skip. Easy to overlook, but its there.
-Original Message-
[mailto:[EMAIL PROTECTED]
Sent: Wednesday, May 28, 2003 11:50 AM
To: Multiple recipients of list ORACLE-L
We have two servers: Server A
Greetings all,
Say I have two main tables in a M:M relationship, and a junction table
resolving the relationship, like so:
create table staff
(staff_id number PRIMARY KEY
, name varchar2(50)
, other cols) ;
create table projects
(proj_id number PRIMARY KEY
, name varchar2(50)
I've done a little bit research and testing on this and one thing that I've
found is that the optimizer will only choose a skip-scan route if the
leading column of the index is relatively non-selective. I haven't been
able to pin down how non-selective it has to be, but I've never had one kick
in
Title: RE: Which method is more efficient
I vote for the whole record, you'd be spending more time and resources in finding _which_ field has changed.
Raj
Rajendra dot Jamadagni at nospamespn dot com
All Views
I was not able to upgrade an 8.1.7.0 database to 9.2.0.1.0 on Win2K. I had
to create a new Oracle home, install 9.2, and clone the database in 9.2.
OUI did not upgrade successfully.
Title: Message
I upgaded 2
productionservers. One running 9 instances, one running 3.
Both are serving data warehouses. Had no problem.
-Original Message-From: Koivu, Lisa
[mailto:[EMAIL PROTECTED] Sent: Wednesday, May 28,
2003 12:25 PMTo: Multiple recipients of list
1. to totally eliminate redo, load your staging records into a global temp table. it
has absolutely no redo and is very fast.
2. if your in 9i, use an external table and a merge command and update the table
directly from the file. dont even load the records to a staging table. Do a search for
Jerry - I did a quick survey of some of my databases and only found a single
table that had a null MAX_EXTENTS and it was not partitioned. Its name was
ATEMPTAB$ owned by SYS.
Dennis Williams
DBA, 80%OCP, 100% DBA
Lifetouch, Inc.
[EMAIL PROTECTED]
-Original Message-
Sent:
Bryan - If this is a critical issue, I would try it both ways on a test
database and use log miner to examine the amount of redo that is generated.
My recollection is that you will find that the redo record records the
before and after data for each field. So just updating all fields may
generate
and THAT is the reason NOT, repeat, NOT to use the OUI to do upgrades. Do
'em manually, I've had a very good success rate that way. Never trust the
GUI's. Never.
Robert
(self-proclaimed GUI biggot)
-Original Message-
To: Multiple recipients of list ORACLE-L
Sent: 5/28/2003 11:56 AM
I
Bryan,
First item is to define efficient. Are you looking for efficiency in the
update of data on the production database or the standby? The code needed to
determine which columns need updating is going to be a real bear and could create
several update statements per record. On the
I've had a couple kick in where full scans were happening before (badly
tuned SQL with out a proper index) and in one case I saw a three index
and-equal become a skip scan.
I'm about 70/30 against skip scans on performance improvements with hints.
RF
-Original Message-
To: Multiple
The execution plan indicates if a skip scan is happening. Can't remember the
exact verbage and I don't have a convienient plan with one handy to pull
out, but you will know it when you see it.
RF
-Original Message-
To: Multiple recipients of list ORACLE-L
Sent: 5/28/2003 10:45 AM
List -
Trifling even if we do an FTS
Until your developers develop a query next month that joins that table to
the 200 million row table they are planning on installing but just forgot to
tell you about.
Developers are funny that way.
Excuse me, did you plan any indexing on this table??
A skip scan can be a index scan, full scan or range scan type access. It
simply allows a unusable column to be deselected from the index (for lack
of a better word) during these operations.
RF
-Original Message-
To: Multiple recipients of list ORACLE-L
Sent: 5/28/2003 11:15 AM
A short
oh i missed part of it. the question is how do you figure out which fields have
changed? if you have to do an anti-join on each field, then do an update of every
field.
the question is how will you determine which fields have changed?
From: DENNIS WILLIAMS [EMAIL PROTECTED]
Date:
Title: Precise Software Solutions
IMHO, Every DBA should own this book. It will change
your way of thinking about how to approach oracleperformance
problems.
Mike
-Original Message-From: Kent Mingus
[mailto:[EMAIL PROTECTED]Sent:
Wednesday, May 28, 2003 6:50 AMTo:
Title: Unbreakable, my buns. Upgrade to 9.2/w2k
I have
upgraded 16 databases ( manually using the scripts and procedures in metablink
as a guide ) with very few problems and only then on the 1st ones I
did.
We
were going from 8174 ro 92021 on Win2k SP3.
-Original Message-From:
As much as I prefer the command line do it yourself approach, I did have
one
occasion recently where I did use the GUI, and it worked fine.
In the process of upgrading several 8.0.4 databases to 8.1.7, I had
trouble with
one database. Sorry, can't recall the details, but I went over the
The past few months I've been wondering when MySQL would start
putting pressure on Oracle in the same way that Linux is putting
pressure on MS.
Maybe sooner than you think:
http://news.com.com/2100-1012_3-1010522.html?tag=fd_top
Jared
--
Please see the official ORACLE-L FAQ:
Actually, it is an index full scan.
Mladen Gogala
Oracle DBA
Phone:(203) 459-6855
Email:[EMAIL PROTECTED]
-Original Message-
Sent: Wednesday, May 28, 2003 12:16 PM
To: Multiple recipients of list ORACLE-L
A short cut to test the new feature is using the hint index_ss(table,index).
I refuse to do business with spammers; regardless of what their product
may be.
Johnson, Michael
Hi dear listers,
Some of you may still remember this thread, bulk collect truncated to 65535
records sometimes.
I've got this case reproduceable and tried all suggestions ,
In a brief,
SELECT returns 318847 rows,
INSERT INTO FROM SELECT - 318847 rows,
PL/SQL plain FOR cr IN (select ..)
Skip scan will show in the execution plan as skip scan. Not true that it
will show as regular index scan.
Waleed
-Original Message-
Sent: Wednesday, May 28, 2003 1:20 PM
To: Multiple recipients of list ORACLE-L
A skip scan can be a index scan, full scan or range scan type access. It
Title: Unbreakable, my buns. Upgrade to 9.2/w2k
Maybe
that's a good opportunity to repost a question before I apply
patchsets:
I actually planned on applying the Patchset to lift 9.2.0.1.0 up to
9.2.0.3.0.
But while downloading, I saw that 9.2.0.3.0 came out on April, 7th, while
there
At http://www.sapdb.org/7.4/pdf/sapdb_letter.pdf SAP offers clarification:
SAP:
Contrary to erroneous press reports, SAP AG has not given up any rights concerning
the SAP DB code base nor handed over or even sold SAP DB to MySQL AB.
SAP:
SAP AG remains responsible for ongoing development and
The fields that are changed are determined by
1) A loop would start until all records in parts change table are done
2) Select a part record from the part changes table
3) Select the same part from the existing part table
4) Compare the value in the parts changes table against the corresponding
Easy test case:
CREATE TABLE TEST_SKIP
(
C1 NUMBER NOT NULL,
C2 NUMBER NOT NULL,
C3 NUMBER NULL
);
CREATE UNIQUE INDEX TESTSKIP1
ON TEST_SKIP(C1,C2);
select --+ index_ss(test_skip, )
c1,c2,c3
from test_skip
where c2 = 10;
OPERATIONOPTIONS
True enough, it will show as index skip scan, but if you take a look at
the statistics, you'll see that the nubmer of blocks read roughly
corresponds
to the number of blocks in the index. It is also logical, because without
the first column, the only way to find the desired key is to read the
Title: Unbreakable, my buns. Upgrade to 9.2/w2k
Hi
Stefan,
I
don't know how much HELP this is, but I've seen several comments on the
list stating that 9.2.0.3 is the version to be at. I'm going to that patch
level from 8.1.7.4. You are right, the docs are confusing.
I've
found that
have you run it? isnt that alot slower? you have alot of context switches also. for
every record to update, you then switch to SQL.
what kind of efficiency improvement are you going for? Speed or cutting down on redo?
Are you in archivelog mode and dont want to blow up your archives?
From:
I am so novice it'll probably take me years to learn oracle
like some of you know it in here...
I am just studying a book, and try to learn from some of its questions.
a. Which util do I use to see oracle's execution plan for a SQL statement?
Which statement would you recommend me, so I can
Reminiscent of the M$/Sybase partnership?
-Original Message-
Sent: Wednesday, May 28, 2003 11:40 AM
To: '[EMAIL PROTECTED]'
At http://www.sapdb.org/7.4/pdf/sapdb_letter.pdf SAP offers clarification:
SAP:
Contrary to erroneous press reports, SAP AG has not given up any rights
All,
Anybody running Oracle Financials on an IBM/AIX platform? I've been asked
for opinions on running the latest version on an IBM P650 platform running
AIX. This would be just for the database itself. Any suggestions for
required memory/disk? I know I need to provide more info (like how
Thanks Waleed. Something even I can understand!
Dennis Williams
DBA, 80%OCP, 100% DBA
Lifetouch, Inc.
[EMAIL PROTECTED]
-Original Message-
Sent: Wednesday, May 28, 2003 1:52 PM
To: Multiple recipients of list ORACLE-L
Easy test case:
CREATE TABLE TEST_SKIP
(
C1 NUMBER NOT NULL,
Thanks everyone for your responses. wish me luck...
Lisa M.
-Original Message-
Sent: Wednesday, May 28, 2003 1:21 PM
To: Multiple recipients of list ORACLE-L
and THAT is the reason NOT, repeat, NOT to use the OUI to do upgrades. Do
'em manually, I've had a very good success rate that
Is it possible that you are running out memory on the OS?
A different question I have is why bulk collect such a large amount at once.
Why not do a cursor with a limit on the fetch? This would allow you to
process in smaller batches instead of one gigantic fetch and insert.
Mark
-Original
I'm talking about the way it get executed not the statistics or the cost.
The cost is completely dependent on the distribution of the data.
For example if we have table (c1 number, c2 number) and a primary key on
(c1, c2).
And the data looks like this:
c1 c2
A 1
A 2
A 3
A 4
. .
.
Title: RE: RE: Which method is more efficient
Bryan,
Can you ...
create table my_work_table as
select * from changed_parts_table
minus
select * from existing_parts_table
/
The result will give you all the rows where _something_ is different between your existing table and changed table.
Hi Thomas,
We are running Oracle Applications Release 11.5.8 on an RS6000 F80 with 4
procs , 8 Gb RAM and 12 x 9.1 Gb , RAID 01 , SSA disks. Both application
and database are on the same machine. The database have almost 50 Gb and
there are aprox. 50 concurrent users , using four modules: GL ,
I'm not talking about the cost either. The way by which is getting executed
is by reading the whole index. You may call it fast full scan, you may call
it index skip scan, but it is still the same thing: sequential read of the
whole index. In other words, the name doesn't matter.
Mladen Gogala
and I had no problem at all doing it. But it was a simple database,
with no major connections or anything.
Maybe because it was a dev box? Oracle didn't think it worthwhile to
mess with my mind?
The only thing that happened was I had to leave the migration of the
database running overnight as it
Interesting, as this book has been out for over a year.
Kirti Deshpande and Gaja Vaidyanatha of this list wrote it
--- Johnson, Michael [EMAIL PROTECTED] wrote:
IMHO, Every DBA should own this book. It will change your way of
thinking
about how to approach oracle performance problems.
this is the online store. If we EVER have a table with more than
100,000 rows in it, I'll faint.
--- Freeman Robert - IL [EMAIL PROTECTED] wrote:
Trifling even if we do an FTS
Until your developers develop a query next month that joins that
table to
the 200 million row table they are
Maryann - No problem, welcome to the list. We only ask that you make a
slight effort to search for the answer first, but these are good questions
that you won't quickly find as a beginner.
a. Oracle provides an explain plan feature to show you how your SQL will be
handled. You must create a table
this is not spam. Michael is signed up to their email listserv and
passed on the information to us, as a member of this list.
how does that become defined as spam?
and oh by the way, Precise has NO connection to Oracle Press or any of
their books
--- [EMAIL PROTECTED] wrote:
I refuse to do
Tom,
Can you clarify what you meant by 'just the database itself'? Fyi, there are
multiple layers in Apps 11i, in which the DB is just one. If this is
*entirely* only the DB (not including any part of the Concurrent
Manager/other apps layers), then the comparison is as if you are hosting
'just a
Rachel, thanks for your response. Did you use the gui?
-Original Message-
Sent: Wednesday, May 28, 2003 4:05 PM
To: Multiple recipients of list ORACLE-L
and I had no problem at all doing it. But it was a simple database,
with no major connections or anything.
Maybe because it was a
Title: RE: Oracle Financials and IBM/AIX Platforms
He's right, Thomas.
We have our middle tier and our Concurrent managers (don't ask) on Win 2K... but our database is on AIX.. and I REALLY wish the rest were too. 50 -80 gig allocated should be fine for the databae layer plus 100 M+- for
I don't think that fainting is in order when you get stinking rich.
If you get more then 100,000 rows in the table, that means that your
company is doing very, very well and that your stock options and your
bonuses will make it possible you to retire to a cosy little place
near the Waikiki Beach.
Could someone give me some information or lead on how I can create a
search engine for usernames. I am running an 8.1.7.4 database and also
9ias for my web application. I have received some info such as to
create the code in ColdFusion and link up to the database with ODBC.
Does anyone have any
Title: RE: RE: Which method is more efficient
And
with CTAS you can specify nologging to minimize redo generation. "Cloning" a
table, renaming/dropping the source, and renaming the clone to the production
table could be interesting. You would have to recreate indexes.
-Original
Not true, try this:
create table test_skip1 ( c1 number,c2 number, primary key (c1,c2));
begin
for i in 1..10 loop
insert into test_skip1 values (1,i);
insert into test_skip1 values (2,i);
end loop;
end;
alter session set sql_trace = true;
select --+ index_ss(test_skip1, )
c1,c2
Mark,
No chance I'm running out of memory. I checked max PGA for the session, it
was around 26M after successful execution, what is not something completely
unbareable, and didn't climb up after next executions.
We can probably workaround this issue, thanks for suggestion.
Regards
Vadim
Roy - This could be a good IOT candidate, but I haven't used them myself. My
suggestion would be to test it. Here is a good paper on how to decide:
http://www.peled.com/white_papers/iot.pdf
A more recent edition is available on http://www.orapub.com (free but must
register)
Dennis Williams
DBA,
Tammy
- For a small amount of searching, take a look at function-based indexes
(you just said usernames).
- You can also create your own method with PL/SQL pretty easily. Strip
individual words into a separate table that can be easily searched.
- For heavy-duty searching, look at Oracle
Tom,
As one who 'transitioned' from a 'normal' DBA (if there ever was a 'normal!)
to an 'Apps DBA', I went through a self-learning path (as did many others in
this list), and tried to capture this transformation in an article. It is
housed at http://www.geocities.com/john_sharmila/links.htm -
Charlie,
I am not a spammer I am a struggling DBA trying to find answers just
like you. I was mired in Hit Ratios for the longest time scratching my
head trying to figure it all out ... lost in the clouds with no way out so
to speak. In the end , it got me no where. I met this
OK. I don't have the 9i instance that I can use for testing right now,
but tonight, at home, I'll give you the counter example. The bottom line
is that the only way to execute a skip scan with a B*Tree index is to
go and read it whole. No other way.
Mladen Gogala
Oracle DBA
Phone:(203) 459-6855
what is this thing you call stock options? My company is already very
rich (Sony) and they don't seem to be passing any of it along to me :(
and it'll be upstate NY or somewhere cold(ish) rather than Waikiki
Beach, I can assure you!
--- Gogala, Mladen [EMAIL PROTECTED] wrote:
I don't think that
You'd see it in the explain plan -- it should look something like this:
TABLE ACCESS (BY INDEX ROWID) OF 'PHONEBOOK'
INDEX (SKIP SCAN) OF 'I_PHONEBOOK_SKIP' (NON-UNIQUE)
--Brian
-Original Message-
Sent: Wednesday, May 28, 2003 9:45 AM
To: Multiple recipients of list
I dunno. Though both want to make a profit ( and rightly so ) SAP
doesn't seem to have the same mercenary mentality that MS has.
Jared
Orr, Steve [EMAIL PROTECTED]
Sent by: [EMAIL PROTECTED]
05/28/2003 11:52 AM
Please respond to ORACLE-L
To: Multiple recipients of list
1 - 100 of 146 matches
Mail list logo