Question

一个人能够举例说明如何在矿石板/板块中使用平行的桌子功能。我们需要在15年内提出大量问题,并将结果结合起来。

SELECT * 
  FROM Table(TableFunction(cursor(SELECT * FROM year_table)))

......是我们有效想要的东西。最廉价的选取将给所有年份,表的功能将逐年进行,并进行大量盘问和交回。我们面临的问题是,所有年份都在坐到一个桌上,我们更希望每年的桌子功能是并行的。我们试图通过洗衣和牧场进行任何形式的分割,而且这样做没有帮助。

而且,我们能否从职能申报中删除关键词? 由于我们没有进行任何转变,因此只需要成果的汇总。

Answer 1

。

还有其他办法(例如,通过YEAR_TABLE进行治疗并提交房舍管理司每年处理的主要工作)。每年的工作都会将成果列入表格。

一旦完成所有 sp的工作,你就只是从桌上提取结果。

PS. I suspect parallel pipelined won t do what you want though. I created a large table with just three rows with a specific value. I then created a parallel pipelined function that just pushed out the SID of the executing process (see below) and the number of rows it processes. I had an SQL that picked out those three rows, and passed that as the the cursor into the function. Mostly the function pushed out two different SIDs (which is what EXPLAIN PLAN told me it picked as the parallelism degree). Sometimes it showed two processes had ed executed, but all three rows were processed by one of those processes.

因此,现在就没有这样的情况,即从 cur子手中夺走row子,然后转到要处理的平行奴隶,但每个平行进程都将获得一张驾驶桌的切片。如果有小的表格,它可能不考虑平行,甚至可能只是把头50分拨到第一程序等。

CREATE OR REPLACE FUNCTION test_pp(p_source     IN SYS_REFCURSOR)
   RETURN TAB_CHAR_4000  PIPELINED
   PARALLEL_ENABLE (PARTITION p_source BY ANY)
IS
   v_num NUMBER;
BEGIN
   FETCH p_source INTO v_num;
   WHILE p_source%FOUND LOOP
            PIPE ROW(sys_context( USERENV , SID ));
            FETCH p_source INTO v_num;
   END LOOP;
     PIPE ROW(sys_context( USERENV , SID )|| : ||p_source%ROWCOUNT);
   CLOSE p_source;
   RETURN;
END test_pp;
/

友情链接