Datawarehousing within Discoverer - Storing snapshots

I'm in the process of constructing a datawarehouse using discoverer.
I need to produce a full set of reports on the data at the end of every month.
I want to keep a full record of what I report every month.
Initially I thought I might be able to archive PDFs or use Scheduled reports.
Each month I will produce a report which includes costs of year to date and a forecast until the end of the year. Once I have made a number of forecasts, I want to compare them with each other and explain exactly why the forecast has changed.
PDF's or Scheduled Reports won't do this because they just store the report but not the underlying data in a form which can further be analysed.
Effectively I want each month to store a snapshot of the whole years data for future reference.
Then as snapshots build up I want to be able to compare snapshots or even show a trend of the results of a snapshot over time. (e.g. a trend of the forecast as well as a trend of expenditure)
Maybe the details of my finances are not of too much importance. What I need is decide the best way to implement a method for storing the whole data situation once a month in such a way that I can reconstuct old reports AND compare old reports.
I wondered about OLAP but am unable to get the license for that.
I also wondered about Materialised views which certainly take a snapshot but as far as I can see when the view is refreshed I lose all the data from the last snapshot.
I'm coming to the conclusion that I need to manually set up a facttable and a number of dimension tables. Then each time I want to create a snapshot, I would copy all the required data from my data staging area, into the fact & dimension tables and include a dimension called "snapshot" which records when I did the snapshot and give it a name (such as "Jun 07"). Then I can keep a record of all snapshots in one model and can report from it AND copare old reports or even show a trend of reports over time.
It's possible I'm missing a feature of DIscoverer that would help?
If anyone knows of anything which would help I'd appreciate you letting me know.
Even if it's to confirm that my proposed approach sounds like the best one.
Kind regards
Suhada

Discoverer is a reporting tool, not one for building a warehouse. The design and implementation of a data warehouse is not the same as the reports that come out of it.
That said, how to solve your problem...
You could use materialized views to store the data monthly, but I think that would be a bit of overkill. One possible solution would be to store all of your data in your fact tables, then create a report in Discoverer that accepts a date as a parameter. The report would calculate/aggregate the data up to that date (from the start of the year), and provide a forecast. The report could then be scheduled to run monthly. If users wanted to see a previous month, they would change the date parameter and rerun the report.
Another problem with the materialized view approach would be late data entry. Suppose there is a scheduled job that creates the MV at midnight on the 1st day of the month. The MV gets created, then the next morning, and employee comes in and finds invoices, orders, or other data that needed to be posted for the previous month. The data could get entered, but the MV is not refreshed, and now the data in the MV is incomplete.

Similar Messages

Creating a table/view or temporary table from within a stored procedure

Hi Gurus,
Can someone tell me if it is possible to create a table (or view) from within a stored procedure.
PROBLEM:
In fact I need to create a report at back end (without using oracle developer forms or reports). This report requires creating several tables to hold temporary report data. If I create a sql*plus script for this, i works fine, because it can run DDL and other sql or pl/sql statements sequencialy. But this sql*plus script cannot be called from application. So, application needs an stored procedure to do this task and then application call that procedure. But within stored procedure, i am unable to create table (or run any ddl statement). Can somebody help me in this?
Thanks in Advance.

Denis,
The problem with Nicholas' suggestion isrelated to the fact that now you have two components
(a table and a stored procedure)
I don't see any problem to have "two
components" here. After all, what about all others
tabes ? This is only one more, but I don't understand
why want manage less objects, that implies more code,
more maintenance, and more difficulties to debug.
Needless to say about performance...
Nicolas.The same reasons apply if you were forced to declare all PL/SQL variables publicly (outside the stored proc.) rather than privately (from inside the stored proc). Naming conflicts for one. If the name that you want to use for the GTT already exists, you need to find a new name. With the SQL Server type local/private declarations, you wouldn't have that problem.
I can see how performance would be the same or better using GTTs. If the number of records involved is low, this is likely negligable.

How to determine the load on the oracle server within a stored procedure?

Hi,
I have a subset of commands in my stored procedure that I would like to execute them only if there is not significant load on the database sever.
What would be the best way to measure the amount of load on my database server within a stored procedure?
Thank you,
Alan

It is not that easy.
This stored procedure is called by a job that runs every hour. Half of the stored procedure code are mission critical and must run by any means necessary! The second half must run right after the 1st one and is nice to do.
If I assign a lower priority it makes the situation worst, because then the job might take 3-4 hours and the same jobs run on top of each other over and over.
I just oversimplified the situation, The point is I prefer not to run the less important PLSQL code rather than assigning lower priority and stretch it over timeline. These hourly jobs are not Oracle jobs and cannot run on top of each other. Last time this happened and I had to kill 12 sessions! (12 jobs in 12 hour)
Trust me, estimating the load on the database server is the only solution.
When I say load on the server I mean the load resulted from other oracle sessions on the same database server. This server is a single instance Oracle 10g Enterprise on Solaris. All I need is just accessing to some meaningful counters.
Any help would be appreciated,
Alan

Sending email from within a stored procedure

Is it possible to send and email within a stored procedure written in PL/SQL. I have searched this site and the web for an Oracle package which simulates a mail agent written in PL/SQL, but I have not had any success.
null

Well, on NT one way we did it was to link to kernel32.dll so we could execute OS commands (WinExec), and did the email that way.
Here's a bit of script to try:
connect sys/@db;
drop library kernellib;
CREATE LIBRARY kernellib as 'c:\winnt\system32\kernel32.dll';
show errors
CREATE OR REPLACE PACKAGE UTIL IS
function run(lpCmdLine IN varchar2, nCmdShow IN binary_integer) return binary_integer;
end;
show errors
CREATE OR REPLACE PACKAGE BODY UTIL IS
function run(lpCmdLine IN varchar2, nCmdShow IN binary_integer) return binary_integer
AS EXTERNAL
NAME "WinExec"
LIBRARY kernellib
parameters (lpCmdLine STRING, nCmdShow long);
end;
show errors
Then you just call util.run() with an appropriate command line to send email. Bit of a hack, but it depends on what you want to do with it. Another hack would be to use the UTL_HTTP package to submit a form to your web server and use a CGI program to send the email.

Connecting to a different schema within a stored procedure

Hi Peopl,
I have a Stored proce "TestProc" which is a part of schema "test1"... Now within this proc i need to connect to a different schema "test2", fetch some data from its tables and disconnect from the same, but not from the original schema "test1". Both the schemas are located on the same database server so i guess no different network connection. I hope u guys got my question... if not tell me... How shud i approach this requirement....

This would all be handled within the same session, so there is no extra connect or disconnect involved here.
The user 'test2' needs to grant the appropriate privileges directly to 'test1'. These grants must be direct and not through a role. If the procedure is just selecting data from test2, then test2 needs to:
test2>grant select on tablea to test1;where tablea represents a table in the test2 schema. There would be a separate grant for each table in test2 accessed by test1.
The code in test1 can refer to the table in test2 as test2.tablea, or you can create a private synonym in test1:
test1>create synonym tablea for test2.tablea;so every reference to tablea does not need to be qualified with the 'test2.' prefix.

Temp Table within an Stored Procedure

I'm pretty new to Oracle, but I have been developing in MS SQL for about 15. So I'm still getting use to the syntax and features within Oracle.
I'm trying to create a stored procedure that has two temporary tables within it, and then queries both them tables and inserts the results into a table.
I created the script but when they try to run in on the server it wont run. Can you guys tell me what I'm doing wrong?
CREATE OR REPLACE PROCEDURE UpdateFIDB_SP IS BEGIN CREATE GLOBAL TEMPORARY TABLE myAAAA AS (SELECT AAAA.1111, AAAA.2222, BBBB.3333_EXT, CCCC.4444, DDDD.5555, DDDD.6666, DDDD.7777, DDDD.8888, AAAA.9999, EEEE.1010, EEEE.1A1A, EEEE.1B1B, FFFF.3333_LO, FFFF.1C1C, AAAA.1D1D FROM mySchema.FFFF_07 FFFF RIGHT OUTER JOIN mySchema.EEEE EEEE ON FFFF.9999 = EEEE.1B1B RIGHT OUTER JOIN ( mySchema.DDDD DDDD RIGHT OUTER JOIN mySchema.AAAA AAAA ON DDDD.1D1D = AAAA.1D1D ) ON EEEE.PSPNR = AAAA.9999 LEFT OUTER JOIN mySchema.CCCC CCCC ON AAAA.3333 = CCCC.3333 LEFT OUTER JOIN mySchema.BBBB BBBB ON AAAA.3333 = BBBB.3333_INT GROUP BY AAAA.1D1D, AAAA.1111, AAAA.2222, BBBB.3333_EXT, CCCC.4444, DDDD.5555, DDDD.6666, DDDD.7777, DDDD.8888, AAAA.9999, EEEE.1010, EEEE.1A1A, EEEE.1B1B, FFFF.3333_LO, FFFF.1C1C ON COMMIT DELETE ROWS); CREATE GLOBAL TEMPORARY TABLE myGGGG AS (SELECT GGGG.1E1E, GGGG.1F1F, GGGG.1G1G, GGGG.1H1H, GGGG.1I1I, GGGG.1J1J, GGGG.1K1K, GGGG.R1D1D, GGGG.1L1L, GGGG.1M1M, GGGG.1N1N, GGGG.1O1O, GGGG.1P1P, GGGG.1Q1Q, HHHH.1R1R, IIII.1S1S, IIII.1T1T, IIII.1U1U, IIII.1V1V FROM mySchema.IIII IIII INNER JOIN mySchema.GGGG GGGG ON IIII.1K1K = GGGG.1K1K LEFT OUTER JOIN mySchema.HHHH HHHH ON GGGG.1L1L = HHHH.1W1W WHERE ( GGGG.1M1M IN ('20', '30') ) AND ( TO_DATE(IIII.1V1V, 'dd-mon-yyyy') = TO_DATE('31-DEC-9999','dd-mon-yyyy') ) AND ( TO_DATE(GGGG.1N1N, 'dd-mon-yyyy') >= TO_DATE('01-Jan-2011','dd-mon-yyyy') ) ON COMMIT DELETE ROWS); TRUNCATE TABLE FIDB; INSERT INTO FIDB (1111, 2222, 3333_EXT, 4444, 5555, 6666, 7777, 8888, 9999, 1010, 1A1A, 1B1B,3333_LO, 1C1C, 1D1D, 1E1E, 1F1F, 1G1G, 1H1H, 1I1I, 1J1J, 1K1K, R1D1D, 1L1L, 1M1M, 1N1N, 1O1O, 1P1P, 1Q1Q, 1R1R, 1S1S, 1T1T, 1U1U, 1V1V) SELECT myAAAA.1111, myAAAA.2222, myAAAA.3333_EXT, myAAAA.4444, myAAAA.5555, myAAAA.6666, myAAAA.7777, myAAAA.8888, myAAAA.9999, myAAAA.1010, myAAAA.1A1A, myAAAA.1B1B, myAAAA.3333_LO, myAAAA.1C1C, myAAAA.1D1D, myGGGG.1E1E, myGGGG.1F1F, myGGGG.1G1G, myGGGG.1H1H, myGGGG.1I1I, myGGGG.1J1J, myGGGG.1K1K, myGGGG.R1D1D, myGGGG.1L1L, myGGGG.1M1M, myGGGG.1N1N, myGGGG.1O1O, myGGGG.1P1P, myGGGG.1Q1Q, myGGGG.1R1R, myGGGG.1S1S, myGGGG.1T1T, myGGGG.1U1U, myGGGG.1V1V FROM myGGGG INNER JOIN myAAAA ON myGGGG.R1D1D = myAAAA.1D1D ORDER BY myGGGG.R1D1D; COMMIT; END;
Thanks!

Some people still sound mad at me because I have two queries and use temp tables
Because in Oracle, unlike in sql server, you generally do not NEED temp tables at all.
One of the biggest mistakes new Oracle developers make is to use PL/SQL when SQL will do the job just fine. Using PL/SQL when it isn't needed generally makes the code slower, less scalable and harder to maintain.
One of the biggest mistakes sql server developers make when they use Oracle is to use temp table when they aren't needed. They try to write code in Oracle exactly the same way they wrote it in sql server. That creates temp tables that aren't needed, makes their code perform horribly and also makes it less scalable and harder to maintain.
Oracle works differently and generally doesn't need ANY temp tables. You need to learn about how Oracle handles transactions and read consistency and how writers do NOT block readers and vice versa.
Create a new thread and post FORMATTED code that shows the PROBLEM you are trying to solve and we can show you the proper way to solve it. That 'proper way' will likely NOT include any need for or use of temp tables.

Is it possible to use iFS API within java stored procedure?

question:
is it possible to use the iFS API for file handling
within a java stored procedure (which is invoced by a trigger)?
is there an "elegant" workaround if it is not possible?
problem:
my java code using the iFS API works fine outside
the java stored procedure
but invoked via trigger as java stored procedure
it throws following exception:
IFS-20102: Unable to start service (IfsDefault)
i am using:
oracle 9.0.1
oracle internet file system 9.0.1.1.0
windows 2000
thanks

thanks for your help
i looked through the applications developers guide
the basics behind advanced queueing (as far as i understand) is:
program A inserts a message into the queue - and program B gets message from the queue and processes it
question 1:
but what is the difference between advanced queuing and a queue i create by myself
(simply with a database table) and let application B (which runs in a loop)
check if the queue is filled
or am i missing something important which makes advanced queuing more practicable for my problem?
question 2:
you wrote about advanced queuing support in iFS 9.0.3, but the newest version i found was iFS 9.0.2
did you reffer to the advanced queuing mechanism in general or to a spezial queuing feature in iFS 9.0.3?
thanks

Colons within document stored as XMLtype

Have uploaded a document in a 10g database, the document has : within the element names. We received the errors
ORA-31011:XML parsing failed
ORA-19202: Error occurred in XML processing
LPX-00601: Invalid token in: '//a:b'
Have discovered that the problem is the : . This worked fine in on an Oracle 9.2 database but now on a 10 release 2 we get this error. Is there any way to work around this without having to change the original xml document

This was my original code to extract the values. I am extracting from a word document saved as xml.
CURSOR c_get_values is
SELECT substr(replace
(replace
(replace
(replace(m.xml_doc_type.extract('//wt').GetClobVal(),'<w:t Xmlns:w="http://schemas.microsoft.com/office/word/2003/wordml">',''),'</wt>',''),
'<w:t xmlns:w="http://schemas.microsoft.com/office/word/2003/wordml"/>',''),chr(10),''),
instr(replace(replace(replace(replace(m.xml_doc_type.extract('//wt').GetClobVal(),
'<w:t xmlns:w="http://schemas.microsoft.com/office/word/2003/wordml">',''),'</wt>',''),'<w:t xmlns:w="http://schemas.microsoft.com/office/word/2003/wordml"/>',''),chr(10),''),v_search1) + length(v_search1),instr(replace(replace(replace(replace(m.xml_doc_type.extract('//wt').GetClobVal(),'<w:t xmlns:w="http://schemas.microsoft.com/office/word/2003/wordml">',''),'</wt>',''),'<w:t xmlns:w="http://schemas.microsoft.com/office/word/2003/wordml"/>',''),chr(10),''),v_search2) - (instr(replace(replace(replace(replace(m.xml_doc_type.extract('//wt').GetClobVal(),'<w:t xmlns:w="http://schemas.microsoft.com/office/word/2003/wordml">',''),'</wt>',''),'<w:t xmlns:w="http://schema--s.microsoft.com/office/word/2003/wordml"/>',''),chr(10),''),v_search1) + length(v_search1)) ) return_value
FROM temp m
WHERE upper(file_name) = UPPER(v_file_name)
AND id = 2 ;
here is an extract from the document.
     <w:tr>
     <w:trPr>
<w:trHeight w:val="567"/>
</w:trPr>
     <w:tc>
     <w:tcPr>
<w:tcW w:w="3338" w:type="dxa"/>
<w:shd w:val="clear" w:color="auto" w:fill="C0C0C0"/>
<w:vAlign w:val="center"/>
</w:tcPr>
     <w:p>
     <w:pPr>
<w:pStyle w:val="Heading2"/>
</w:pPr>
     <w:r>
<w:t>USER_LOCATION</w:t>
</w:r>
</w:p>
</w:tc>
     <w:tc>
     <w:tcPr>
<w:tcW w:w="6721" w:type="dxa"/>
<w:shd w:val="clear" w:color="auto" w:fill="FFFFFF"/>
<w:vAlign w:val="center"/>
</w:tcPr>
     <w:p>
     <w:pPr>
     <w:rPr>
<w:rFonts w:ascii="Arial" w:h-ansi="Arial" w:cs="Arial"/>
<wx:font wx:val="Arial"/>
<w:sz w:val="20"/>
<w:sz-cs w:val="20"/>
</w:rPr>
</w:pPr>
     <st1:City w:st="on">
     <st1:place w:st="on">
     <w:r>
     <w:rPr>
<w:rFonts w:ascii="Arial" w:h-ansi="Arial" w:cs="Arial"/>
<wx:font wx:val="Arial"/>
<w:sz w:val="20"/>
<w:sz-cs w:val="20"/>
</w:rPr>
<w:t>Edinburgh</w:t>
</w:r>
</st1:place>
</st1:City>
</w:p>
</w:tc>
</w:tr>
I have amended the code to extract the clob, then amend the data to remove the : and then insert into a table as an XMLType. Would be better if didn't have to do this.

SELECT query from within a stored procedure

DB Version:10gR2
For auditing purpose, i just want to execute the query within the IF condition, and get the results for spooling. But i am getting PLS-00428 error. Can't i just execute a SELECT query and get its results displayed in SQLPLUS so that i can spool them?
SQL> create or replace procedure ship_dtl_aud
2 is
3 v_exists number;
4
5 begin
6
7 Select count(1) into v_exists
8 From ship_dtl
9 WHERE track_code = 10
10 AND mod_date_time < SYSDATE - 1/(24*60);
11
12   if v_exists>0
13   then
14            select s.username,s.sid, s.serial#, s.terminal,p.spid
15   from v$session s, v$process p
16   where s.paddr = p.addr and s.sid in (select SESSION_ID from V$LOCKED_OBJECT);
17
18          end if;
19 end;
20 /
Warning: Procedure created with compilation errors.
SQL> show errors
Errors for PROCEDURE ship_dtl_aud:
LINE/COL ERROR
14/11    PLS-00428: an INTO clause is expected in this SELECT statement

Within PL/SQL code, the results of a query have to be returned from the SQL engine to the PL/SQL engine, hence why it is asking you to include an INTO within your query. Note that the PL/SQL engine is running on the database server so it has no interface to display output to. To output from within PL/SQL you would have to write the data out yourself to whatever output you want e.g. use DBMS_OUTPUT to put the data to the standard output buffer (which can then be read by SQL*Plus if serveroutput is set to "on") or written out to a file on the server, or written out to a table.
You are assuming that because you can execute SQL from SQL*Plus and spool the output, you can do the same within PL/SQL. This is not the case. When SQL*Plus issues a SQL statement, it goes to the SQL engine and the results of that cursor (as an SQL query is a cursor) are collected by SQL*Plus and it displays them itself. When PL/SQL issues a SQL statement, it goes to the SQL engine and the results of that cursor come back to PL/SQL. PL/SQL doesn't have a display mechanism, so the results have to be collected INTO something.
Hope that's clear.

Using Java within a stored proc to access multiple database connections

Hi List;
We have an environment where we have needs to provide cross database visibility between Oracle, Sql Server and PostgreSQL. So we need the capability for Oracle to query PostgreSQL and Sql Server tables, PostgreSQL must be able to query Oracle and Sql Server tables and Sql Server must be able to query Oracle and PostgreSQL tables.
I'm thinking that we can implement stored procs which are either written in java or call a java process . The java code in turn should be able to connect to any DBMS desired via JDBC query the tables in question and return the result set to the calling database.
Any thoughts on how feasable this is would be appreciated. Does this seem like a solid solution? Any initial concerns or red flags? what about performance via JDBC?
Thanks in advance for your help.

John,
from your description I understand that you essentially program the VO yourself. So I suggest that you read chapter 35.9 of the 'Fusion Developer’s Guide for Oracle Application Development Framework' (I guess you know where to find it). Since you still call the actual SQL not all of the chapter apply, but you get the idea how it works.
And yes you analyzed the behavior correct. executeQueryForCollection() is allways called bevore getQueryHitCount().
Timo

Running "Host" From Within a Stored Procedure

Hellow
Wanna Run a "Host" Command from Within A SP
or DB Trigger
Any Help is Appreciated !!
Regards
Tariq

the only way to run an OS command from within the database is through extproc
(works from 8.x onwards).
Essentially, you have to:
- modify the database listener to support external procedure calls
- create a library object
- create a C - program and compile it as a shared library
- write a wrapper function in PL/SQL that calls the shared library
I can send you an example, if you want.
Regards
Anton Weindl

Where is IDENT(Item ID) within BOM stored

Does any one now where the item id(IDENT) field value really stored in DB..
It is not stored STOP..I am wondering if it would be stored in different table..
Any help would be appreciated.
Thanks,
Frank

Rich,
How can we determine STLKN..?if you want item id (STKLN) depending on STLNR and IDNRK every time you hit the same number...
Am I right..?or else is there another key field where I should I take into account..?
Any help would be appreciated.
Thanks,
Frank.

Script to copy all the objects within a Stored Procedure

I am trying to copy all the objects from one database to another using Scripts. Now I need to put that script inside a Stored Procedure.
Thanks.

Refer this vbscript to generate all object scripts programatically,
http://www.codeproject.com/Articles/16195/How-to-Script-Out-SQL-Server-Objects
or
Using SSMS
http://msdn.microsoft.com/en-IN/library/hh245282.aspx#GenPubScriptWiz
Regards, RSingh

Oracle SID from within a stored proc

Sorry - probably a no-brainer, but can't find it anywhere in the doc. Can anyone tell me how to query the current session identifier in pl/sql. (ie. the SID I can use to find an entry in v$$session.
Thanks,Tim

http://asktom.oracle.com/pls/ask/f?p=4950:8:::::F4950_P8_DISPLAYID:114412348062

How to use dbms_Scheduler.Create_Job from within stored procedure?

Hello,
using 10g (10.1.0.2.0) on Windows 2000 I had problems to create scheduler jobs from within a stored procedure (see example below). What easily succeeds using anonymous blocks failed when calling from a stored procedure, due to ORA-27486. Only when I compile the procedure with invoker's rights the call to dbms_Job.Create_Job is successfull!? From my knowledge there is no difference between invoker's and definer's rights, if I compile and call with the same user, is there?
Does anyone know the reason for this behaviour or is it simply a bug?
Have a nice day.
BjÃ¶rn Hachmann
Hamburg / Germany
-- Example start.
create table t
a number(1),
b date default sysdate
create or replace procedure sched1
is
begin
dbms_scheduler.create_job(
job_name => 'TEST_JOB1',
job_type => 'PLSQL_BLOCK',
job_action => 'BEGIN insert into t values (1); END;',
repeat_interval => 'freq=secondly',
enabled => TRUE
commit;
end;
create or replace procedure sched2
authid current_user
is
begin
dbms_scheduler.create_job(
job_name => 'TEST_JOB2',
job_type => 'PLSQL_BLOCK',
job_action => 'BEGIN insert into t values (2); END;',
repeat_interval => 'freq=secondly',
enabled => TRUE
commit;
end;
exec sched1; -- This call fails with ORA-27486.
exec sched2; -- This call succeeds!
/* Cleanup.
exec dbms_scheduler.drop_job('TEST_JOB1', true);
exec dbms_scheduler.drop_job('TEST_JOB2', true);
drop table t;
*/

Your example code ran without problems for me on 10.1.0.3.0 so it probably is a bug.

Datawarehousing within Discoverer - Storing snapshots

Similar Messages

Maybe you are looking for