Monster Oasis

覺得累就是進步的開始!

• mysql 的 innodb 重裝或改了 innodb_log_file_size 後, 發現 xxx/yyy.frm 壞了 , 解決辦法是 把 /var/lib/mysql/ib_logfile* 砍了, 再 restart mysql
• 在 http://dev.mysql.com/doc/refman/5.1/en/alter-table.html 中提到….

  若要大量 bluk 作 insert 動作前, 下 ALTER TABLE tbl_name DISABLE KEYS , 這樣可以讓 insert 加快,

  但是作完 insert 後還是得 enable keys , 把 missing 的 indexs 補回來, 我想這時也是非常耗時間吧!!

  另外 enable / disable keys 對於 mysql 5.1.1 以前的 partition table 沒有用

參考: http://dev.mysql.com/doc/refman/5.0/en/optimizing-database-structure.html

7.4.1. Make Your Data as Small as Possible
7.4.2. Column Indexes
7.4.3. Multiple-Column Indexes
7.4.4. How MySQL Uses Indexes
7.4.5. The MyISAM Key Cache
7.4.6. MyISAM Index Statistics Collection
7.4.7. How MySQL Opens and Closes Tables
7.4.8. Disadvantages of Creating Many Tables in the Same Database

減少 record structure 的大小 – Numeric Types 參考表
(用 MEDIUMINT 3bytes 比 INT 4bytes 好 , 若資料內容不可能有負值那加上 UNSIGNED , 則數值範圍可多一倍!)

http://dev.mysql.com/doc/refman/5.0/en/multiple-column-indexes.html

CREATE TABLE test (
    id         INT NOT NULL,
    last_name  CHAR(30) NOT NULL,
    first_name CHAR(30) NOT NULL,
    PRIMARY KEY (id),
    INDEX name (last_name,first_name)
);

The name index is an index over the last_name and first_name columns. The index can be used for queries that specify values in a known range for last_name, or for both last_name and first_name. Therefore, the name index is used in the following queries:

SELECT * FROM test WHERE last_name='Widenius';

SELECT * FROM test
  WHERE last_name='Widenius' AND first_name='Michael';

SELECT * FROM test
  WHERE last_name='Widenius'
  AND (first_name='Michael' OR first_name='Monty');

SELECT * FROM test
  WHERE last_name='Widenius'
  AND first_name >='M' AND first_name < 'N';

However, the name index is not used in the following queries, 以下的 query 用不到 index —> 多重 index 有先後區別.

SELECT * FROM test WHERE first_name='Michael';

SELECT * FROM test
  WHERE last_name='Widenius' OR first_name='Michael';

http://www.dba-oracle.com/t_sql_like_clause_index_usage.htm

Indexing when using the SQL "like" clause can be tricky because the wildcard "%" operator can invalidate the index.  For example a last_name index would be OK with a "like ‘SMI%’" query, but unusable with "like ‘%SMI%’.

Solutions to this issue of a leading wildcard can be addressed in several ways::

• Oracle text indexes to remove full-table scans when using the LIKE operator.

Burleson Consulting 說:

These unnecessary full-table scans are a problem:
1. Large-table full-table scans increase the load on the disk I/O sub-system

2. Small table full table scans (in the data buffer) cause high consistent gets and drive-up CPU consumption

非必要的 full-table-scans 造成幾個問題 : 大的資料表會增加 disk I/O , 小的資料表則是增加 CPU 消耗. 所以這個現象可以拿來觀察資料庫系統的問題點.

Tuning Oracle Full-table Scans

http://www.orafaq.com/node/39

常見Oracle HINT的用法：

1. /*+ALL_ROWS*/
表明對語句塊選擇基於開銷的優化方法,並獲得最佳吞吐量,使資源消耗最小化.
例如:
SELECT /*+ALL+_ROWS*/ EMP_NO,EMP_NAM,DAT_IN FROM BSEMPMS WHERE EMP_NO=’SCOTT’;
2. /*+FIRST_ROWS*/
表明對語句塊選擇基於開銷的優化方法,並獲得最佳響應時間,使資源消耗最小化.
例如:
SELECT /*+FIRST_ROWS*/ EMP_NO,EMP_NAM,DAT_IN FROM BSEMPMS WHERE EMP_NO=’SCOTT’;

3. /*+CHOOSE*/
表明如果數據字典中有訪問表的統計資料,將基於開銷的優化方法,並獲得最佳的吞吐量;
表明如果數據字典中沒有訪問表的統計資料,將基於規則開銷的優化方法;
例如:
SELECT /*+CHOOSE*/ EMP_NO,EMP_NAM,DAT_IN FROM BSEMPMS WHERE EMP_NO=’SCOTT’;

4. /*+RULE*/
表明對語句塊選擇基於規則的優化方法.
例如:
SELECT /*+ RULE */ EMP_NO,EMP_NAM,DAT_IN FROM BSEMPMS WHERE EMP_NO=’SCOTT’;

5. /*+FULL(TABLE)*/
表明對表選擇全局掃描的方法.
例如:
SELECT /*+FULL(A)*/ EMP_NO,EMP_NAM FROM BSEMPMS A WHERE EMP_NO=’SCOTT’;

6. /*+ROWID(TABLE)*/
提示明確表明對指定表根據ROWID進行訪問.
例如:
SELECT /*+ROWID(BSEMPMS)*/ * FROM BSEMPMS WHERE ROWID>=’AAAAAAAAAAAAAA’
AND EMP_NO=’SCOTT’;

7. /*+CLUSTER(TABLE)*/
提示明確表明對指定表選擇簇掃描的訪問方法,它只對簇對像有效.
例如:
SELECT /*+CLUSTER */ BSEMPMS.EMP_NO,DPT_NO FROM BSEMPMS,BSDPTMS
WHERE DPT_NO=’TEC304′ AND BSEMPMS.DPT_NO=BSDPTMS.DPT_NO;

8. /*+INDEX(TABLE INDEX_NAME)*/
表明對錶選擇索引的掃描方法.
例如:
SELECT /*+INDEX(BSEMPMS SEX_INDEX) USE SEX_INDEX BECAUSE THERE ARE FEWMALE BSEMPMS */ FROM BSEMPMS WHERE SEX=’M';

9. /*+INDEX_ASC(TABLE INDEX_NAME)*/
表明對錶選擇索引升序的掃描方法.
例如:
SELECT /*+INDEX_ASC(BSEMPMS PK_BSEMPMS) */ FROM BSEMPMS WHERE DPT_NO=’SCOTT’;

10. /*+INDEX_COMBINE*/
為指定表選擇位圖訪問路經,如果INDEX_COMBINE中沒有提供作為參數的索引,將選擇出位圖索引的布爾組合方式.
例如:
SELECT /*+INDEX_COMBINE(BSEMPMS SAL_BMI HIREDATE_BMI)*/ * FROM BSEMPMS
WHERE SAL<5000000 emp_no="SCOTT" sex="M" dpt_no="V.DPT_NO">V.AVG_SAL;

20. /*+NO_MERGE(TABLE)*/
對於有可合併的視圖不再合併.
例如:
SELECT /*+NO_MERGE(V) */ A.EMP_NO,A.EMP_NAM,B.DPT_NO FROM BSEMPMS A (SELECT DPT_NO,AVG(SAL) AS AVG_SAL FROM BSEMPMS B GROUP BY DPT_NO) V WHERE A.DPT_NO=V.DPT_NO AND A.SAL>V.AVG_SAL;

21. /*+ORDERED*/
根據表出現在FROM中的順序,ORDERED使ORACLE依此順序對其連接.
例如:
SELECT /*+ORDERED*/ A.COL1,B.COL2,C.COL3 FROM TABLE1 A,TABLE2 B,TABLE3 C WHERE A.COL1=B.COL1 AND B.COL1=C.COL1;

22. /*+USE_NL(TABLE)*/
將指定表與嵌套的連接的行源進行連接,並把指定表作為內部表.
例如:
SELECT /*+ORDERED USE_NL(BSEMPMS)*/ BSDPTMS.DPT_NO,BSEMPMS.EMP_NO,BSEMPMS.EMP_NAM FROM BSEMPMS,BSDPTMS WHERE BSEMPMS.DPT_NO=BSDPTMS.DPT_NO;

23. /*+USE_MERGE(TABLE)*/
將指定的表與其他行源通過合併排序連接方式連接起來.
例如:
SELECT /*+USE_MERGE(BSEMPMS,BSDPTMS)*/ * FROM BSEMPMS,BSDPTMS WHERE BSEMPMS.DPT_NO=BSDPTMS.DPT_NO;

24. /*+USE_HASH(TABLE)*/
將指定的表與其他行源通過hash連接方式連接起來.
例如:
SELECT /*+USE_HASH(BSEMPMS,BSDPTMS)*/ * FROM BSEMPMS,BSDPTMS WHERE BSEMPMS.DPT_NO=BSDPTMS.DPT_NO;

25. /*+DRIVING_SITE(TABLE)*/
強制與ORACLE所選擇的位置不同的表進行查詢執行.
例如:
SELECT /*+DRIVING_SITE(DEPT)*/ * FROM BSEMPMS,DEPT@BSDPTMS WHERE BSEMPMS.DPT_NO=DEPT.DPT_NO;

26. /*+LEADING(TABLE)*/
將指定的表作為連接次序中的首表.

27. /*+CACHE(TABLE)*/
當進行 full scan 時,CACHE提示能夠將表的檢索塊放置在緩衝區緩存中最近最少列表LRU的最近使用端
例如:
SELECT /*+FULL(BSEMPMS) CAHE(BSEMPMS) */ EMP_NAM FROM BSEMPMS;

28. /*+NOCACHE(TABLE)*/
當進行 full scan 時,CACHE提示能夠將表的檢索塊放置在緩衝區緩存中最近最少列表LRU的最近使用端
例如:
SELECT /*+FULL(BSEMPMS) NOCAHE(BSEMPMS) */ EMP_NAM FROM BSEMPMS;

29. /*+APPEND*/
直接插入到表的最後,可以提高速度.
insert /*+append*/ into test1 select * from test4 ;

30. /*+NOAPPEND*/
通過在插入語句生存期內停止並行模式來啟動常規插入.
insert /*+noappend*/ into test1 select * from test4 ;

可以用
Using Per-Table Tablespaces : http://dev.mysql.com/doc/refman/5.1/en/multiple-tablespaces.html

Store each InnoDB table and its indexes in its own file. This feature is called “multiple tablespaces” because in effect each table has its own tablespace.

Using multiple tablespaces can be beneficial to users who want to move specific tables to separate physical disks or who wish to restore backups of single tables quickly without interrupting the use of other InnoDB tables.

[mysqld]
innodb_file_per_table

那在把不同的 innodb file 放到不同 disk partition 去, 達到分散實體 IO 的目標, 不過, mysql 對於 path 部份要特別處理, 避免 drop table 或 alter table 時, 它會發生無法預期的結果, 翻了 web , 看到有人這麼作:

改 mysys/mf_format.c

if (!lstat(to,&stat_buff))  /* Check if it's a symbolic link */
    if (S_ISLNK(stat_buff.st_mode) && realpath(to,buff))
成

if (realpath(to,buff))

試看看吧, 我沒改過!