It shows a number of useful metrics for CPU, memory, disk, network etc…if you use nmon, but didn’t know it was available for linux, well, now you do.

I’d always used the ROW_NUMBER method myself until I read this and then figured I’d try the KEEP DENSE_RANK method, which works, as Laurent describes. One thing that didn’t sit well with me in the post from Laurent was that he said “the second one should be more performant” – I prefer hard facts, so I decided to test it a bit and my results are below.

In the simple example that Laurent gave, it’s difficult to tell which is quickest, since the table in question only has a handful of rows and therefore any benchmarking is more susceptible to other influences, clouding the results. I figured I’d build a larger table and try it on that.

Before I did that though, I did get the plans from the two statements Laurent ran on the EMP table and both show the same resource costings:

Firstly, for the ROW_NUMBER method:

select ename
,      deptno
,      sal
from   (select ename
 ,      deptno
 ,      sal
 ,      row_number() over (partition by deptno order by sal desc,empno) r
 from   emp
 )
where  r=1;

Plan hash value: 3291446077                                                                                                             

---------------------------------------------------------------------------------
| Id  | Operation                | Name | Rows  | Bytes | Cost (%CPU)| Time     |
---------------------------------------------------------------------------------
|   0 | SELECT STATEMENT         |      |    14 |   644 |     4  (25)| 00:00:01 |
|*  1 |  VIEW                    |      |    14 |   644 |     4  (25)| 00:00:01 |
|*  2 |   WINDOW SORT PUSHED RANK|      |    14 |   644 |     4  (25)| 00:00:01 |
|   3 |    TABLE ACCESS FULL     | EMP  |    14 |   644 |     3   (0)| 00:00:01 |
---------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

 1 - filter("R"=1)
 2 - filter(ROW_NUMBER() OVER ( PARTITION BY "DEPTNO" ORDER BY
 INTERNAL_FUNCTION("SAL") DESC ,"EMPNO")<=1)

Note
-----
 - dynamic sampling used for this statement

Now, the KEEP DENSE_RANK method:

select max(ename) keep (dense_rank first order by sal desc,empno) ename
,      deptno
,      max(sal) sal
from   emp 
group by deptno;

Plan hash value: 15469362

---------------------------------------------------------------------------
| Id  | Operation          | Name | Rows  | Bytes | Cost (%CPU)| Time     |
---------------------------------------------------------------------------
|   0 | SELECT STATEMENT   |      |    14 |   644 |     4  (25)| 00:00:01 |
|   1 |  SORT GROUP BY     |      |    14 |   644 |     4  (25)| 00:00:01 |
|   2 |   TABLE ACCESS FULL| EMP  |    14 |   644 |     3   (0)| 00:00:01 |
---------------------------------------------------------------------------

Note
-----
   - dynamic sampling used for this statement

So, the plans are very similar, especially in terms of having the same resource usage…which means they should be similar in terms of performance…running them, as Laurent did, shows around 1s response times, which, as I say, doesn’t conclusively prove which method is quickest.

OK, on to a bigger example then…

I basically created a similar table to EMP, called JEFF_EMP and added a few more columns (for later) and then put ten million rows in it, taking around 1.3GB on my system…plenty to churn through.

DROP TABLE jeff_emp PURGE
/
CREATE TABLE jeff_emp(deptno     NUMBER
                     ,ename      VARCHAR2(100)
                     ,first_name VARCHAR2(50)
                     ,initials   VARCHAR2(30)
                     ,surname    VARCHAR2(50)
                     ,sal        NUMBER
                     ,empno      NUMBER
                     )
/
INSERT INTO jeff_emp(deptno,ename,first_name,initials,surname,sal,empno)
SELECT (MOD(ROWNUM,3) + 1) * 10
,      'FIRSTNAME_'||TO_CHAR(ROWNUM)||'_INITIALS_'||TO_CHAR(ROWNUM)||'_SURNAME_'||TO_CHAR(ROWNUM)
,      'FIRSTNAME_'||TO_CHAR(ROWNUM)
,      'INITIALS_'||TO_CHAR(ROWNUM)
,      'SURNAME_'||TO_CHAR(ROWNUM)
,      ROWNUM * 100
,      ROWNUM
FROM   (SELECT LEVEL l FROM dual CONNECT BY LEVEL < 10000001)
ORDER BY l
/
COMMIT
/
EXEC dbms_stats.gather_table_stats(ownname => USER, tabname => 'JEFF_EMP',estimate_percent=>10);

Now, here is the plan for the ROW_NUMBER method:

---------------------------------------------------------------------------------------------
| Id  | Operation                | Name     | Rows  | Bytes |TempSpc| Cost (%CPU)| Time     |
---------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT         |          |    10M|   868M|       |   204K  (1)| 00:40:49 |
|*  1 |  VIEW                    |          |    10M|   868M|       |   204K  (1)| 00:40:49 |
|*  2 |   WINDOW SORT PUSHED RANK|          |    10M|   629M|  1533M|   204K  (1)| 00:40:49 |
|   3 |    TABLE ACCESS FULL     | JEFF_EMP |    10M|   629M|       | 46605   (1)| 00:09:20 |
---------------------------------------------------------------------------------------------

…and the results:

ENAME                                                                                                    DEPTNO        SAL
---------------------------------------------------------------------------------------------------- ---------- ----------
FIRSTNAME_9999999_INITIALS_9999999_SURNAME_9999999                                                           10  999999900
FIRSTNAME_10000000_INITIALS_10000000_SURNAME_10000000                                                        20 1000000000
FIRSTNAME_9999998_INITIALS_9999998_SURNAME_9999998                                                           30  999999800

Elapsed: 00:00:24.47

…and the KEEP DENSE_RANK method plan:

-------------------------------------------------------------------------------
| Id  | Operation          | Name     | Rows  | Bytes | Cost (%CPU)| Time     |
-------------------------------------------------------------------------------
|   0 | SELECT STATEMENT   |          |     3 |   198 | 47109   (2)| 00:09:26 |
|   1 |  SORT GROUP BY     |          |     3 |   198 | 47109   (2)| 00:09:26 |
|   2 |   TABLE ACCESS FULL| JEFF_EMP |    10M|   629M| 46605   (1)| 00:09:20 |
-------------------------------------------------------------------------------

…and it’s results:

ENAME                                                                                                    DEPTNO        SAL
---------------------------------------------------------------------------------------------------- ---------- ----------
FIRSTNAME_9999999_INITIALS_9999999_SURNAME_9999999                                                           10  999999900
FIRSTNAME_10000000_INITIALS_10000000_SURNAME_10000000                                                        20 1000000000
FIRSTNAME_9999998_INITIALS_9999998_SURNAME_9999998                                                           30  999999800

Elapsed: 00:00:07.76

So, reasonably clear results, indicating that the KEEP DENSE_RANK is about a third of the time to run, compared to the ROW_NUMBER method. You can also see from the plans that the ROW_NUMBER method involves use of TEMP, whereas the KEEP DENSE_RANK doesn’t, hence the slowdown.

So, Laurent was correct in his assertion that it should be more performant…but it’s nice to see the results based on a more meaningful set of data.

Now, there was one other thing that concerned me, and that was whether if you added more columns into the SQL, would it change the performance fo either method to any significant degree, so I started using the extra name columns like this:

SELECT ename
,      first_name
,      initials
,      surname
,      deptno
,      sal
FROM   (SELECT ename
        ,      first_name
        ,      initials
        ,      surname
        ,      deptno
        ,      sal
        ,      ROW_NUMBER() OVER (PARTITION BY deptno ORDER BY sal DESC,empno) r 
        FROM   jeff_emp
       )
WHERE  r = 1
/

…which has a plan of:

---------------------------------------------------------------------------------------------
| Id  | Operation                | Name     | Rows  | Bytes |TempSpc| Cost (%CPU)| Time     |
---------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT         |          |    10M|  1546M|       |   307K  (1)| 01:01:36 |
|*  1 |  VIEW                    |          |    10M|  1546M|       |   307K  (1)| 01:01:36 |
|*  2 |   WINDOW SORT PUSHED RANK|          |    10M|  1107M|  2606M|   307K  (1)| 01:01:36 |
|   3 |    TABLE ACCESS FULL     | JEFF_EMP |    10M|  1107M|       | 46605   (1)| 00:09:20 |
---------------------------------------------------------------------------------------------

…and results:

ENAME                                                                                                FIRST_NAME                                         INITIALS                       SURNAME                      DEPTNO         SAL
---------------------------------------------------------------------------------------------------- -------------------------------------------------- ------------------------------ -------------------------------------------------- ---------- ----------
FIRSTNAME_9999999_INITIALS_9999999_SURNAME_9999999                                                   FIRSTNAME_9999999                                  INITIALS_9999999               SURNAME_9999999          10  999999900
FIRSTNAME_10000000_INITIALS_10000000_SURNAME_10000000                                                FIRSTNAME_10000000                                 INITIALS_10000000              SURNAME_10000000         20 1000000000
FIRSTNAME_9999998_INITIALS_9999998_SURNAME_9999998                                                   FIRSTNAME_9999998                                  INITIALS_9999998               SURNAME_9999998          30  999999800

Elapsed: 00:00:25.76

For the KEEP DENSE_RANK I get:

SELECT MAX(ename) KEEP (DENSE_RANK FIRST ORDER BY sal DESC,empno) ename
,      MAX(first_name) KEEP (DENSE_RANK FIRST ORDER BY sal DESC,empno) first_name
,      MAX(initials) KEEP (DENSE_RANK FIRST ORDER BY sal DESC,empno) initials
,      MAX(surname) KEEP (DENSE_RANK FIRST ORDER BY sal DESC,empno) surname
,      deptno
,      MAX(sal) sal
FROM   jeff_emp 
GROUP BY deptno
/

Which has the following plan:

-------------------------------------------------------------------------------
| Id  | Operation          | Name     | Rows  | Bytes | Cost (%CPU)| Time     |
-------------------------------------------------------------------------------
|   0 | SELECT STATEMENT   |          |     3 |   348 | 47109   (2)| 00:09:26 |
|   1 |  SORT GROUP BY     |          |     3 |   348 | 47109   (2)| 00:09:26 |
|   2 |   TABLE ACCESS FULL| JEFF_EMP |    10M|  1107M| 46605   (1)| 00:09:20 |
-------------------------------------------------------------------------------

…and results:

ENAME                                                                                                FIRST_NAME                                         INITIALS                       SURNAME                      DEPTNO         SAL
---------------------------------------------------------------------------------------------------- -------------------------------------------------- ------------------------------ -------------------------------------------------- ---------- ----------
FIRSTNAME_9999999_INITIALS_9999999_SURNAME_9999999                                                   FIRSTNAME_9999999                                  INITIALS_9999999               SURNAME_9999999          10  999999900
FIRSTNAME_10000000_INITIALS_10000000_SURNAME_10000000                                                FIRSTNAME_10000000                                 INITIALS_10000000              SURNAME_10000000         20 1000000000
FIRSTNAME_9999998_INITIALS_9999998_SURNAME_9999998                                                   FIRSTNAME_9999998                                  INITIALS_9999998               SURNAME_9999998          30  999999800

Elapsed: 00:00:14.56

So, the differential in performance has reduced significantly, with the KEEP DENSE_RANK around double it’s original time, whilst the ROW_NUMBER method has only increased marginally. I’ve not tested with adding additional columns, but I’m guessing (I know…I could and should test it!) it will get worse, to the extent that, eventually, the KEEP DENSE_RANK will become the worse performer. If that’s the case, then essentially, these two methods have different scalability dynamics and one should bear this in mind when considering which to choose, depending on how many DENSE_RANK’d columns you’d need to deliver your results.

Hope this helps.

RANK, ROW_NUMBER and also DENSE_RANK are very useful for taking a set of rows and ordering them in a defined manner, whilst giving each row a “position value”. They differ based on the approach taken to define the value of their position in the set of output rows. In some circumstances, they may all give the same value however, dependent on the data, they may differ.

An example based on the SCOTT.EMP table, helps to illustrate…

SELECT empno
,      sal
,      RANK() OVER(ORDER BY sal) rank_position
,      DENSE_RANK() OVER(ORDER BY sal) dense_rank_position
,      ROW_NUMBER() OVER(ORDER BY sal) row_number_position
FROM   emp
/

which returns, on my 11gR1 database…

EMPNO        SAL RANK_POSITION DENSE_RANK_POSITION ROW_NUMBER_POSITION
---------- ---------- ------------- ------------------- -------------------
7369        800             1                   1                   1
7900        950             2                   2                   2
7876       1100             3                   3                   3
7521       1250             4                   4                   4
7654       1250             4                   4                   5
7934       1300             6                   5                   6
7844       1500             7                   6                   7
7499       1600             8                   7                   8
7782       2450             9                   8                   9
7698       2850            10                   9                  10
7566       2975            11                  10                  11
7788       3000            12                  11                  12
7902       3000            12                  11                  13
7839       5000            14                  12                  14
14 rows selected.

Notice that RANK has given the two employees with SAL = 1250, the same position value of 4 and the two employees with SAL=3000, the same position value of 12. Notice also that RANK skips position values 5 and 13 as it has two entries for 4 and 12 respectively. RANK uses all numbers between 1 and 14, except 5 and 13. RANK has both repeats and gaps in it’s ordering.

DENSE_RANK is similar to RANK, in that it gives the two employees with SAL=1250, the same position value of 4, but then it does not skip over position value 5 – it simply carries on at position 5 for the next values. DENSE_RANK uses, for the position values, all numbers between 1 and 12, without leaving any out, and using 4 and 11 twice. DENSE_RANK has no gaps in it’s ordering, only repeats.

ROW_NUMBER gives each row a unique position value and consequently uses all the numbers between 1 and 14. ROW_NUMBER has no gaps or repeats in it’s ordering. Note that the position value on ROW_NUMBER is not deterministic, since the ORDER BY clause only has SAL in it. If you want to ensure the order is the same each time, you need to add further columns to the ORDER BY clause.

I started looking at the first Oracle By Example OBIEE tutorial yesterday and came across an issue with a simple solution to share with you.

The tutorials have a few caveats about what they expect from your environment, when you’re going to run through them – one of them being that you have access to a 10g database. What the prerequisites don’t specifically say is that unless your database is called ORCL, you’ll need to jump through a few more hoops – as I did, with my database called TEST.

I followed the instructions from the tutorial such that I had:

1. An SH schema in the 10g database, with the standard tables and data Oracle supply.
2. Created an ODBC Data Source pointing to a database called TEST, checking it functioned correctly.
3. Restored the presentation catalog and updated the configuration files accordingly.

I then proceeded to login to the BI Dashboards which brought up half the display, but it was full of TNS errors indicating that the connection could not be made between the BI Server and the database where the SH schema resides:

After some investigation in the Administration tool, I discovered that the Connection Pool setting was using a Data Source Name called “ORCL” which doesn’t match my TNS/Database called TEST, hence it couldn’t make the connection to the database:

Now, the RPD was read only at the time, so I first shut down the services so it could be opened read/write:

…logged into the Administration tool using Administrator user (Password Administrator), opened the SH.RPD file read/write and modified the Data Source Name in the Connection Pool from ORCL to TEST, whilst ensuring the password for the SH user matched that of my TEST database:

…next I restarted the services:

…and then logged on again, to find it all now worked:

I did a simple select from DBA_TAB_PARTITIONS and noticed that some tablespace_names were of the form “_$deleted$n$m” where n and m are numbers. Slightly worrying, but at least the data was all present and correct, when I checked. I knew the DBA team had been doing some reorganisations the previous weekend, to recover some space, so I wondered if that was connected….it was, and after opening an SR, the DBA, Phil, found an explanation (from Oracle Note: 604648.1) and a resolution.

Reproducing the issue and the way to fix it, is simple, using this script…

DROP TABLESPACE old_tbs INCLUDING CONTENTS AND DATAFILES;

CREATE TABLESPACE new_tbs 
DATAFILE 'C:APPORACLEORADATAT111NEW_TBS.DBF'
SIZE 100M
ONLINE;

CREATE TABLESPACE old_tbs 
DATAFILE 'C:APPORACLEORADATAT111OLD_TBS.DBF'
SIZE 100M
ONLINE;

SELECT ts#,name FROM sys.ts$ WHERE name LIKE '%TBS';

CREATE TABLE jeff_test(col1 DATE NOT NULL
                      ,col2 NUMBER NOT NULL
                      ,col3 VARCHAR2(200) NOT NULL
                      )
TABLESPACE old_tbs
PARTITION BY RANGE(col1)
SUBPARTITION BY LIST(col2)
SUBPARTITION TEMPLATE(
 SUBPARTITION "S1" VALUES(1)
,SUBPARTITION "S2" VALUES(2)
)
(PARTITION p1 VALUES LESS THAN(TO_DATE('31-DEC-2009','DD-MON-YYYY'))
,PARTITION p2 VALUES LESS THAN(TO_DATE('31-DEC-2010','DD-MON-YYYY'))
)
/

SELECT partition_name,tablespace_name FROM dba_tab_partitions WHERE table_name='JEFF_TEST';
SELECT subpartition_name,tablespace_name FROM dba_tab_subpartitions WHERE table_name='JEFF_TEST';

ALTER TABLE jeff_test MOVE SUBPARTITION p1_s1 TABLESPACE NEW_TBS;
ALTER TABLE jeff_test MOVE SUBPARTITION p1_s2 TABLESPACE NEW_TBS;
ALTER TABLE jeff_test MOVE SUBPARTITION p2_s1 TABLESPACE NEW_TBS;
ALTER TABLE jeff_test MOVE SUBPARTITION p2_s2 TABLESPACE NEW_TBS;

DROP TABLESPACE old_tbs INCLUDING CONTENTS AND DATAFILES;
ALTER TABLESPACE new_tbs RENAME TO old_tbs;

SELECT partition_name,tablespace_name FROM dba_tab_partitions WHERE table_name='JEFF_TEST';
SELECT subpartition_name,tablespace_name FROM dba_tab_subpartitions WHERE table_name='JEFF_TEST';

ALTER TABLE jeff_test MODIFY DEFAULT ATTRIBUTES FOR PARTITION p1 TABLESPACE old_tbs;
ALTER TABLE jeff_test MODIFY DEFAULT ATTRIBUTES FOR PARTITION p2 TABLESPACE old_tbs;

SELECT partition_name,tablespace_name FROM dba_tab_partitions WHERE table_name='JEFF_TEST';
SELECT subpartition_name,tablespace_name FROM dba_tab_subpartitions WHERE table_name='JEFF_TEST';

Which, when run in 11.1.0.6 on Windows 2003 Server 64 bit, gives:

DROP TABLESPACE old_tbs succeeded.
CREATE TABLESPACE succeeded.
CREATE TABLESPACE succeeded.
TS#                    NAME                           
---------------------- ------------------------------ 
9                      NEW_TBS                        
10                     OLD_TBS                        

2 rows selected

CREATE TABLE succeeded.
PARTITION_NAME                 TABLESPACE_NAME                
------------------------------ ------------------------------ 
P1                             OLD_TBS                        
P2                             OLD_TBS                        

2 rows selected

SUBPARTITION_NAME              TABLESPACE_NAME                
------------------------------ ------------------------------ 
P1_S2                          OLD_TBS                        
P1_S1                          OLD_TBS                        
P2_S2                          OLD_TBS                        
P2_S1                          OLD_TBS                        

4 rows selected

ALTER TABLE jeff_test succeeded.
ALTER TABLE jeff_test succeeded.
ALTER TABLE jeff_test succeeded.
ALTER TABLE jeff_test succeeded.
DROP TABLESPACE old_tbs succeeded.
ALTER TABLESPACE new_tbs succeeded.
PARTITION_NAME                 TABLESPACE_NAME                
------------------------------ ------------------------------ 
P1                             _$deleted$10$0                 
P2                             _$deleted$10$0                 

2 rows selected

SUBPARTITION_NAME              TABLESPACE_NAME                
------------------------------ ------------------------------ 
P1_S2                          OLD_TBS                        
P1_S1                          OLD_TBS                        
P2_S2                          OLD_TBS                        
P2_S1                          OLD_TBS                        

4 rows selected

ALTER TABLE jeff_test succeeded.
ALTER TABLE jeff_test succeeded.
PARTITION_NAME                 TABLESPACE_NAME                
------------------------------ ------------------------------ 
P1                             OLD_TBS                        
P2                             OLD_TBS                        

2 rows selected

SUBPARTITION_NAME              TABLESPACE_NAME                
------------------------------ ------------------------------ 
P1_S2                          OLD_TBS                        
P1_S1                          OLD_TBS                        
P2_S2                          OLD_TBS                        
P2_S1                          OLD_TBS                        

4 rows selected

Notice that the $n in “_$deleted$n$m” is 10, which is the ts# of the OLD_TBS before the rename. The problem revolves around entries in TS$, when you rename tablespaces to names that have previously been used and then dropped, basically because the old entries are not removed from TS$.

Related references:
Bug Numbers:8291493, itself a duplicate of 5769963
Note: 604648.1

According to the SR and bug, it was noticed in 10.2.0.4 and is fixed in 10.2.0.5. We’ve reproduced it in 11.1.0.6 on various ports, (results above) and updated our SR, so I guess the fix might also find it’s way into 11.1.0.7, perhaps.

NOTE – I tried this on 10.2.04 (AIX) and 11.1.0 (Fedora 11) – the example below is from 11.1.0.

I’ll recreate the scenario here:

CREATE TABLESPACE tsp1
datafile '/u01/app/oracle/oradata/T111/tsp1.dbf' size 100M 
autoextend off extent management local  uniform size 1m segment space management auto online
/
CREATE TABLESPACE tsp2
datafile '/u01/app/oracle/oradata/T111/tsp2.dbf' size 100M 
autoextend off extent management local  uniform size 1m segment space management auto online
/

DROP TABLE test PURGE
/
CREATE TABLE test(col_date_part_key DATE            NOT NULL
                 ,col2              VARCHAR2(2000)  NOT NULL
                 )
PARTITION BY RANGE(col_date_part_key)
(PARTITION month_01 VALUES LESS THAN (TO_DATE('01-FEB-2009','DD-MON-YYYY')) TABLESPACE tsp1
,PARTITION month_02 VALUES LESS THAN (TO_DATE('01-MAR-2009','DD-MON-YYYY')) TABLESPACE tsp2
,PARTITION month_03 VALUES LESS THAN (TO_DATE('01-APR-2009','DD-MON-YYYY')) TABLESPACE tsp2
,PARTITION month_04 VALUES LESS THAN (TO_DATE('01-MAY-2009','DD-MON-YYYY')) TABLESPACE tsp2
,PARTITION month_05 VALUES LESS THAN (TO_DATE('01-JUN-2009','DD-MON-YYYY')) TABLESPACE tsp2
,PARTITION month_06 VALUES LESS THAN (TO_DATE('01-JUL-2009','DD-MON-YYYY')) TABLESPACE tsp2
,PARTITION month_07 VALUES LESS THAN (TO_DATE('01-AUG-2009','DD-MON-YYYY')) TABLESPACE tsp2
,PARTITION month_08 VALUES LESS THAN (TO_DATE('01-SEP-2009','DD-MON-YYYY')) TABLESPACE tsp2
,PARTITION month_09 VALUES LESS THAN (TO_DATE('01-OCT-2009','DD-MON-YYYY')) TABLESPACE tsp2
,PARTITION month_10 VALUES LESS THAN (TO_DATE('01-NOV-2009','DD-MON-YYYY')) TABLESPACE tsp2
,PARTITION month_11 VALUES LESS THAN (TO_DATE('01-DEC-2009','DD-MON-YYYY')) TABLESPACE tsp2
,PARTITION month_12 VALUES LESS THAN (TO_DATE('01-JAN-2010','DD-MON-YYYY')) TABLESPACE tsp2
)
/
REM Insert rows, but only up to 14-JUN-2009
INSERT INTO test(col_date_part_key,col2)
SELECT TO_DATE('31-DEC-2008','DD-MON-YYYY') + l
,      LPAD('X',2000,'X')
FROM   (SELECT level l FROM dual CONNECT BY level < 166)
/
COMMIT
/
SELECT COUNT(*)
FROM   test
/
SELECT MIN(col_date_part_key) min_date
,      MAX(col_date_part_key) max_date
FROM   test
/

This runs and gives the following output:

DROP TABLE test PURGE                                               
           *                                                        
ERROR at line 1:                                                    
ORA-00942: table or view does not exist                             


DROP TABLESPACE tsp1 INCLUDING CONTENTS
*                                      
ERROR at line 1:                       
ORA-00959: tablespace 'TSP1' does not exist


DROP TABLESPACE tsp2 INCLUDING CONTENTS
*                                      
ERROR at line 1:
ORA-00959: tablespace 'TSP2' does not exist



Tablespace created.


Tablespace created.


Table created.


165 rows created.


Commit complete.


  COUNT(*)
----------
       165


MIN_DATE  MAX_DATE
--------- ---------
01-JAN-09 14-JUN-09

Now, lets see what the plan looks like from AUTOTRACE when we run the following query to get the maximum value of COL_DATE_PART_KEY:

SQL> SET AUTOTRACE ON
SQL> SELECT MAX(col_date_part_key) min_date
  2  FROM   test                           
  3  /                                     

MIN_DATE
---------
14-JUN-09


Execution Plan
----------------------------------------------------------
Plan hash value: 784602781                                

---------------------------------------------------------------------------------------------
| Id  | Operation            | Name | Rows  | Bytes | Cost (%CPU)| Time     | Pstart| Pstop |
---------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT     |      |     1 |     9 |    99   (0)| 00:00:02 |       |       |
|   1 |  SORT AGGREGATE      |      |     1 |     9 |            |          |       |       |
|   2 |   PARTITION RANGE ALL|      |   132 |  1188 |    99   (0)| 00:00:02 |     1 |    12 |
|   3 |    TABLE ACCESS FULL | TEST |   132 |  1188 |    99   (0)| 00:00:02 |     1 |    12 |
---------------------------------------------------------------------------------------------

Note
-----
   - dynamic sampling used for this statement


Statistics
----------------------------------------------------------
          0  recursive calls
          0  db block gets
        320  consistent gets
         51  physical reads
          0  redo size
        527  bytes sent via SQL*Net to client
        524  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          0  sorts (memory)
          0  sorts (disk)
          1  rows processed

SQL> SET AUTOTRACE OFF

It shows a full scan of all twelve partitions. I figured that the the plan for such a query would show a full table scan, of all partitions for that table – because, in theory, if all but the first partition were empty, then the whole table would have to be scanned to answer the query – and Oracle wouldn’t know at plan creation time, whether the data met this case, so it would have to do the full table scan to ensure the correct result.

What I thought might happen though, is that in executing the query, it would be able to short circuit things, by working through the partitions in order, from latest to earliest, and finding the first, non null, value. Once it found the first, non null, value, it would know not to continue looking in the earlier partitions, since the value of COL_DATE_PART_KEY couldn’t possibly be greater than the non null value already identified.

It doesn’t appear to have this capability, which we can check by taking one of the partitions offline and then rerunning the query, whereupon it complains that not all the data is present…

SQL> ALTER TABLESPACE tsp1 OFFLINE;

Tablespace altered.

SQL> SET AUTOTRACE ON
SQL> SELECT MAX(col_date_part_key) min_date
  2  FROM   test
  3  /
SELECT MAX(col_date_part_key) min_date
*
ERROR at line 1:
ORA-00376: file 6 cannot be read at this time
ORA-01110: data file 6: '/u01/app/oracle/oradata/T111/tsp1.dbf'


SQL> SET AUTOTRACE OFF

So, even though we know we could actually answer this question accurately, Oracle can’t do it as it wants to scan, unnecessarily, the whole table.

I did find a thread which somebody had asked about this on OTN, but all the responses were about workarounds, rather than explaining why this happens (bug/feature) or how it can be made to work in the way I, or the poster of that thread, think it, perhaps, should.

Can anyone else shed any light on this? If it’s a feature, then it seems like something that could be easily coded more efficiently by Oracle. The same issue would affect both MIN and MAX since both could be
approached in the same manner.

The installation I undertook was on a FC10 64 bit VM running under VMWare Server 2.0 running on top of FC10 64 bit OS.

Tim, as usual, has a lovely guide which told me almost everything I needed to know, however the guide says “If you are performing the 64-bit installation, make sure both the 32-bit and 64-bit libraries are installed.” rather than explicitly stating the packages for a 64 bit install. Until I tried to install Oracle 11gR1, I didn’t know what these were. The Oracle installer for 11g soon told me in the pre install checks it does, so I went about installing the following packages, in order:

glibc-2.9-3.i686.rpm
libaio-0.3.107-4.fc10.i386.rpm
libgcc-4.3.2-7.i386.rpm
glibc-devel-2.9-3.i386.rpm
compat-libstdc++-33-3.2.3-64.x86_64.rpm
compat-libstdc++-33-3.2.3-64.i386.rpm
libstdc++-4.3.2-7.i386.rpm

That got me past the pre install checks of the Oracle installer and on to a successful install.

I’ve added the list to the comments on the guide Tim produced as well.

I couldn’t get some of the cursor keys to work properly on my virtual machines running under VMWare Virtual Server 2 on Fedora 10 x86_64. Kept giving funny behaviour like bringing up the screen capture applet!

A bit of searching the net came up with this one, which although not referring to Virtual Server 2 specifically, seems to work all the same…

Essentially, adding the line below to the following file fixes the problem

File (create it, if not already present):

~/.vmware/config

Line:

xkeymap.nokeycodeMap = true

My thanks to “The Monkey Jungle”!

NOTE – I don’t have access to other version of Windows Server (2003 or 2008) so I can’t speak for them, but I imagine it’s the same on them too.

That’s annoying for home use, where I have tons of VMs for research that I use periodically, so I asked my brother Steve how to stop this happening and he gave me some simple instructions…

First start the Local Security Policy editor by typing secpol.msc in the start/run box..

and then select Account Policies / Password Policy on the nagivation tree on the left. On the right hand side select Maximum Password Age and set this from the default of “42″ to “0″. You’ll notice it now says “Password will not expire” above the value “0″.

Seems to have done the trick.

Kinda handy having a brother who’s an MCSE and a VCP too. Useful when I get stuck with OS or VM stuff for my Oracle research!

By the way, if anyone happens to be looking for some skilled contract resource in the Virtualisation field (VMWare, ESX Server etc…) then Steve has just become available…please feel free to contact me, or Steve, via his website