Parallel DML in 12c => /*+ parallel enable_parallel_dml */

-
Eine sehr gute Erklärung zum PARALLEL Insert ab 12.1.0.2 findet sind hier:

Quelle: https://blog.dbi-services.com/parallel-dml-in-12c/


aus dem Post:
-----------------------------------------------------------------------------------------------
| Id  | Operation                | Name     | Starts | E-Rows | A-Rows |   A-Time   | Buffers |
-----------------------------------------------------------------------------------------------
|   0 | INSERT STATEMENT         |          |      1 |        |      0 |00:00:17.40 |   24311 |
|   1 |  LOAD TABLE CONVENTIONAL | DEMO2    |      1 |        |      0 |00:00:17.40 |   24311 |
|   2 |   PX COORDINATOR         |          |      1 |        |   1000K|00:00:04.49 |       5 |
|   3 |    PX SEND QC (RANDOM)   | :TQ10000 |      0 |   1000K|      0 |00:00:00.01 |       0 |
|   4 |     PX BLOCK ITERATOR    |          |      0 |   1000K|      0 |00:00:00.01 |       0 |
|*  5 |      TABLE ACCESS FULL   | DEMO1    |      0 |   1000K|      0 |00:00:00.01 |       0 |
-----------------------------------------------------------------------------------------------
 
Note                                                                                                                               
-----
   - Degree of Parallelism is 2 because of table property
   - PDML is disabled in current session

=> The select part is done in parallel (it’s below the coordinator) but the insert part (LOAD TABLE) is above the coordinator, which means that it is done in serial by the coordinator.


Beispiel 2
insert /*+ parallel enable_parallel_dml */ into DEMO2 select * from DEMO1
 
Plan hash value: 86785878
 
-------------------------------------------------------------------------------------------------------
| Id  | Operation                          | Name | Starts | A-Rows | Buffers | OMem |1Mem | Used-Mem |
-------------------------------------------------------------------------------------------------------
|   0 | INSERT STATEMENT                   |      |      1 |      4 |     136 |      |     |          |
|   1 |  PX COORDINATOR                    |      |      1 |      4 |     136 |      |     |          |
|   2 |   PX SEND QC (RANDOM)              | :TQ10|      0 |      0 |       0 |      |     |          |
|   3 |    LOAD AS SELECT (HYBRID TSM/HWMB)|      |      0 |      0 |       0 |   33M|  33M| 2068K (0)|
|   4 |     OPTIMIZER STATISTICS GATHERING |      |      0 |      0 |       0 |      |     |          |
|   5 |      PX BLOCK ITERATOR             |      |      0 |      0 |       0 |      |     |          |
|*  6 |       TABLE ACCESS FULL            | DEMO1|      0 |      0 |       0 |      |     |          |
-------------------------------------------------------------------------------------------------------
 
Note
-----
   - automatic DOP: Computed Degree of Parallelism is 2

=> Here PDML occurred. We know that because of the load operator under the coordinator







Kommentare

Kommentar veröffentlichen

Beliebte Posts aus diesem Blog

PGA unter Oracle 11g

-
aus der Doku: Quelle: http://download.oracle.com/docs/cd/E11882_01/server.112/e16638/memory.htm#PFGRF94351 7.5 PGA Memory Management For complex queries (for example, decision support queries), a big portion of the run-time area is dedicated to work areas allocated by memory intensive operators, such as the following: * Sort-based operators, such as ORDER BY, GROUP BY, ROLLUP, and window functions * Hash-join * Bitmap merge * Bitmap create * Write buffers used by bulk load operations Sizing: A) Ideally, the size of a work area is big enough that it can accommodate the input data and auxiliary memory structures allocated by its associated SQL operator. This is known as the optimal size of a work area. B) When the size of the work area is smaller than optimal, the response time increases, because an extra pass is performed over part of the input data. This is known as the one-pass size of the work area. C) Under the one-pass threshold, when the size of a work area is far too small compar
Mehr anzeigen

trunc(sysdate) - nette Spiele mit dem Datum

-
Über die trunc-date Funktion läßt sich ausgehend von einem Datumswert sehr einfach ein abgeleiteter Datumswert ermittelt: bspw. der Jahres-, Quartals-, Monats- oder Wochenanfang. Die 11g Doku erläutert dies an zwei Stellen: http://download.oracle.com/docs/cd/B28359_01/server.111/b28286/functions209.htm#SQLRF06151 bzw. http://download.oracle.com/docs/cd/B28359_01/server.111/b28286/functions242.htm#sthref2718 So liefert select trunc(to_date('06.02.2011'),'q') from dual; das DATUM des Quartalsersten (im Bsp. den 01.01.2011). Wichtig ist, dass trunc immer ein Datum zurückliefert, keinen String. Einige Beispiele anhand des 20.Feb 2011 (ein Sonntag in der 07 KW in 2011): select trunc(to_date('20.02.2011'),'d') from dual; select trunc(to_date('20.02.2011'),'dd') from dual; select trunc(to_date('20.02.2011'),'ddd') from dual; select trunc(to_date('20.02.2011'),'mm') from dual; select trunc(to_date('20.02.2011&
Mehr anzeigen

Datapump - Verzeichnis erstellen

-
siehe auch Metalink-Doc: 266875.1 HOW TO SPECIFY A DIRECTORY IN ORACLE10G EXPORT/IMPORT DATAPUMP -------------------------------------------------------------- 1. SERVER-BASED OR CLIENT-BASED ---------------------------- The parameter DIRECTORY specifies the location to which Export DataPump or Import DataPump can write the dump file set, the log file, and the SQL file (Import DataPump only). As export DataPump and import DataPump are server-based, rather than client-based, the output files are accessed relative to server-based directory paths. Data Pump requires you to specify directory paths as directory objects. A directory object maps a name to a directory path on the file system. 2. HOW TO CREATE A DIRECTORY ------------------------- To create a directory, you must have the DBA role or you must have been granted the CREATE ANY DIRECTORY privilege. Example (a DBA creates directories on the Windows platform and grants access to user scott): SQL> CONNECT
Mehr anzeigen