我编写了一个视图逻辑来根据要求提取数据。我需要建议或解决方案来提高视图逻辑的性能。
表:
|SID | SREF | STYPE | SQID | SNAME | CCID1 | CCID2 | CCID3 |
--------------------------------------------------------------------
|XX | SREF001 | HR | 1 | 10001 | ABC | DEF | XYZ |
|XX | SREF001 | HR | 2 | 10001 | ABC | DEF | XYZ |
|XX | SREF002 | KR | 1 | 10001 | ABC | DEF | XYZ |
|XX | SREF002 | KR | 2 | 10001 | ABC | DEF | XYZ |
|YY | SREF011 | HR | 1 | 20001 | ABC | DEF | XYZ |
|YY | SREF011 | HR | 2 | 20001 | ABC | DEF | XYZ |
|YY | SREF012 | KR | 1 | 20001 | ABC | DEF | XYZ |
|YY | SREF012 | KR | 2 | 20001 | ABC | DEF | XYZ |
我的观点:
CREATE VIEW MYTAB_VIEW AS
SELECT SID, SREF,
MAX(CASE WHEN STYPE='HR' AND SQID=1 THEN SNAME END) AS HR1_SNAME,
MAX(CASE WHEN STYPE='HR' AND SQID=1 THEN CCID1 END) AS HR1_CCID1,
MAX(CASE WHEN STYPE='HR' AND SQID=1 THEN CCID2 END) AS HR1_CCID2,
MAX(CASE WHEN STYPE='HR' AND SQID=1 THEN CCID3 END) AS HR1_CCID3,
MAX(CASE WHEN STYPE='HR' AND SQID=2 THEN SNAME END) AS HR2_SNAME,
MAX(CASE WHEN STYPE='HR' AND SQID=2 THEN CCID1 END) AS HR2_CCID1,
MAX(CASE WHEN STYPE='HR' AND SQID=2 THEN CCID2 END) AS HR2_CCID2,
MAX(CASE WHEN STYPE='HR' AND SQID=2 THEN CCID3 END) AS HR2_CCID3,
MAX(CASE WHEN STYPE='KR' AND SQID=1 THEN SNAME END) AS KR1_SNAME,
MAX(CASE WHEN STYPE='KR' AND SQID=1 THEN CCID1 END) AS KR1_CCID1,
MAX(CASE WHEN STYPE='KR' AND SQID=1 THEN CCID2 END) AS KR1_CCID2,
MAX(CASE WHEN STYPE='KR' AND SQID=1 THEN CCID3 END) AS KR1_CCID3,
MAX(CASE WHEN STYPE='KR' AND SQID=2 THEN SNAME END) AS KR2_SNAME,
MAX(CASE WHEN STYPE='KR' AND SQID=2 THEN CCID1 END) AS KR2_CCID1,
MAX(CASE WHEN STYPE='KR' AND SQID=2 THEN CCID2 END) AS KR2_CCID2,
MAX(CASE WHEN STYPE='KR' AND SQID=2 THEN CCID3 END) AS KR2_CCID3
FROM MYTABLE
GROUP BY SID, SREF;
“MYTABLE”将有数百万的记录。因此,视图逻辑成为生产中最大的性能打击。我有 STYPE、SQID、SNAME、CCID1、CCID2、CCID3 的索引。但是还是很慢...
执行时间需要 105 秒,现在!!
请建议我如何提高性能...
更新:请不要建议MaterializedView
。
谢谢!