为了更好的测试我们的系统在一定数据量下的运行情况，时常需要准备大量的测试数据。如果有灵活的方法可以使用，那就可以事半功倍了。
以下介绍如何在ORACLE数据库中进行大数据量的构造方法，数据量大小均为500万。
'''基本主键列'''
SELECT LEVEL AS ID
FROM DUAL
CONNECT BY LEVEL <= 5000000
'''基本的单据列'''
SELECT 'SO20121123' || LPAD(LEVEL, 5, 0) AS ORDER_NO
FROM DUAL
CONNECT BY LEVEL <= 5000000;
'''基本的日期列'''
SELECT TO_DATE(TRUNC(DBMS_RANDOM.VALUE(TO_NUMBER(TO_CHAR(SYSDATE, 'J')),
TO_NUMBER(TO_CHAR(SYSDATE, 'J')))),
'J') + DBMS_RANDOM.VALUE(1, 360000) / 3600 AS ENTRYDATE
FROM DUAL
CONNECT BY LEVEL < 5000000;
'''特定时间范围内'''
SELECT TO_DATE(TRUNC(DBMS_RANDOM.VALUE(TO_NUMBER(TO_CHAR(SYSDATE - 10, 'J')),
TO_NUMBER(TO_CHAR(SYSDATE, 'J')))),
'J') + DBMS_RANDOM.VALUE(1, 360000) / 3600 / 24 AS ENTRYDATE
FROM DUAL
CONNECT BY LEVEL < 5000000;
'''随机数据'''
SELECT DBMS_RANDOM.VALUE FROM DUAL CONNECT BY LEVEL < 5000000;
'''随机数据范围'''
SELECT TRUNC(DBMS_RANDOM.VALUE(0, 100))
FROM DUAL
CONNECT BY LEVEL < 5000000;
'''随机字符串'''
SELECT DBMS_RANDOM.STRING('A', 8) FROM DUAL CONNECT BY LEVEL < 5000000;
'''复杂组合（订单）'''
SELECT 'P-' ||
TO_CHAR((TO_DATE(TRUNC(DBMS_RANDOM.VALUE(TO_NUMBER(TO_CHAR(SYSDATE - 30,
'J')),
TO_NUMBER(TO_CHAR(SYSDATE,
'J')))),
'J') + DBMS_RANDOM.VALUE(1, 360000) / 3600 / 24),
'YYMMDD') || '-' ||
LPAD(TRUNC(DBMS_RANDOM.VALUE(1, 100)), 5, '0') AS 订单号,
'一般采购' AS 订单类型,
'SO20121123' || LPAD(LEVEL, 4, 0) AS 外部订单号,
'admin' AS 变更人,
TO_DATE(TRUNC(DBMS_RANDOM.VALUE(TO_NUMBER(TO_CHAR(SYSDATE, 'J')),
TO_NUMBER(TO_CHAR(SYSDATE, 'J')))),
'J') + DBMS_RANDOM.VALUE(1, 360000) / 3600 AS 更新时间
FROM DUAL
CONNECT BY LEVEL < 5000000;
'''加入随机用户或者字符'''
SELECT 'P-' ||
TO_CHAR((TO_DATE(TRUNC(DBMS_RANDOM.VALUE(TO_NUMBER(TO_CHAR(SYSDATE - 30,
'J')),
TO_NUMBER(TO_CHAR(SYSDATE,
'J')))),
'J') + DBMS_RANDOM.VALUE(1, 360000) / 3600 / 24),
'YYMMDD') || '-' ||
LPAD(TRUNC(DBMS_RANDOM.VALUE(1, 100)), 5, '0') AS 订单号,
'一般采购' AS 订单类型,
CASE TRUNC(DBMS_RANDOM.VALUE(0, 2))
WHEN 0 THEN
('SO20121123' || LPAD(LEVEL, 4, 0))
ELSE
('')
END AS 外部订单号,
CASE TRUNC(DBMS_RANDOM.VALUE(0, 3))
WHEN 0 THEN
'admin'
WHEN 1 THEN
'steven'
ELSE
'joseph'
END AS 变更人,
TO_DATE(TRUNC(DBMS_RANDOM.VALUE(TO_NUMBER(TO_CHAR(SYSDATE, 'J')),
TO_NUMBER(TO_CHAR(SYSDATE, 'J')))),
'J') + DBMS_RANDOM.VALUE(1, 360000) / 3600 AS 更新时间
FROM DUAL
CONNECT BY LEVEL < 5000000;