Question

我想知道是否有更好或更好的方式来编写Bulk Collects。我已经看到了不同的方法，其中一些在下面的示例代码中：

set serveroutput on;
declare

    cursor  ext_data_file is 
    select  *
    from    PER_ALL_PEOPLE_F;

    TYPE    type_emp_rec_ext IS TABLE OF ext_data_file%ROWTYPE index by pls_integer;

    l_emp_tab_ext_raw         type_emp_rec_ext;
    l_emp_tab_ext_clean       type_emp_rec_ext;   
    l_start                   number;
    l_count                   number := 0;  
    l_count2                  number := 0;  

begin

    /* Regular Cursor Fetch */

    l_start := DBMS_UTILITY.get_time;

    FOR cur_rec IN ext_data_file LOOP
        l_count2 := l_count2 + 1;
    END LOOP;

    DBMS_OUTPUT.put_line('Regular    (' || l_count2 || ' rows): ' || (DBMS_UTILITY.get_time - l_start) ||' ms');

    /* Bulk Collect 1 */

    l_start := DBMS_UTILITY.get_time;

    open   ext_data_file;
    fetch  ext_data_file
    bulk collect
    into   l_emp_tab_ext_raw;
    close  ext_data_file;

    DBMS_OUTPUT.put_line('Bulk 1  (' || l_emp_tab_ext_raw.count || ' rows): ' || (DBMS_UTILITY.get_time - l_start) ||' ms');

    /* Bulk Collect 2 */

    l_start := DBMS_UTILITY.get_time;

    SELECT *
    BULK COLLECT INTO 
    l_emp_tab_ext_clean
    FROM   PER_ALL_PEOPLE_F;

    DBMS_OUTPUT.put_line('Bulk 2  (' || l_emp_tab_ext_raw.count || ' rows): ' || (DBMS_UTILITY.get_time - l_start) ||' ms');

    /* Bulk Collect 3 */

    l_start := DBMS_UTILITY.get_time;

    OPEN ext_data_file;
    LOOP
        FETCH ext_data_file
        BULK COLLECT INTO l_emp_tab_ext_raw LIMIT 1000;    
        l_count := l_emp_tab_ext_raw.count + l_count;
        EXIT WHEN l_emp_tab_ext_raw.count = 0;        
    END LOOP;
    CLOSE ext_data_file;

    DBMS_OUTPUT.put_line('Bulk 3  (' || l_count || ' rows): ' || (DBMS_UTILITY.get_time - l_start) ||' ms');

EXCEPTION
    WHEN OTHERS THEN
        dbms_output.put_line(sqlerrm);
end;

在表现方面，我惊讶地发现存在差异。常规提取比批量收集更快！

Regular (1202666 rows): 4174 ms
Bulk 1  (1202666 rows): 6369 ms
Bulk 2  (1202666 rows): 7204 ms
Bulk 3  (1202666 rows): 4380 ms

以下是我的观察：

Bulk Collect 1 : Commonly used when using an Explicit Cursor and has no LIMIT Clause. Also when the cursor Involves numerous Joined Tables.
Bulk Collect 2 : Commonly used when using a single table or an Implicit Cursor and has no LIMIT Clause.
Bulk Collect 3 : Commonly used when using an Explicit Cursor and has a LIMIT Clause.

请确认以上观察结果是否正确。

我的问题是：

1. Why is the Regular Fetch Faster?
2. Which of these Bulk Collects are the Best Syntax to Use in Terms of Overall Performance and Memory Consumption)

我一直在使用语法1，但我认为最好的是语法3。

数据库详细信息：

Oracle Database 11g Enterprise Edition Release 11.2.0.4.0 - 64bit Production
PL/SQL Release 11.2.0.4.0 - Production
CORE    11.2.0.4.0  Production
TNS for Solaris: Version 11.2.0.4.0 - Production
NLSRTL Version 11.2.0.4.0 - Production

谢谢！

Answer 1

为什么没有批量收集会更快？因为您对从光标获取的数据不执行任何操作。通常，一个人在更新/插入数据时使用批量收集来批量处理它，而不是逐行处理。尝试在循环中执行，将这些行插入到另一个表中，然后您就会看到差异。

哪种方式更好？这取决于您的要求。对我而言，最有意义的解决方案是限制。如果我要像1和2那样处理整套数据，为什么要批量收集？这可能意味着我能够在一个庞大的声明中处理数据。如果使用限制，您可以优化批次大小以获得最佳的处理数据性能。此外，限制有助于在插入速度和锁定表的时间之间找到折衷方式，以便进行插入/更新。

批量收集的最佳语法

1 个答案: