【性能优化】之 BITMAP 及分区表 的演示
以下演示有两个重点：
1.BITMAP索引的局限性： DML操作导致位图索引锁定；
2.分区索引的性能并没有比全局索引更优。
书面作业，必须给出全部的演示过程。<br>
1.分别给出一个B-tree索引针对全表扫描性能高和低的例子。<br>
2.分别给出一个Bitmap索引针对b-tree索引性能高和低的例子。<br>
3.演示DML操作导致位图索引锁定示例。<br>
4.创建一个全文索引，比较它和传统的模糊查询的性能。<br>
5.分别演示分区索引的性能优化全局索引和差于全局索引的示例，并分析原因。<br>
====================================================================================
1.分别给出一个B-tree索引针对全表扫描性能高和低的例子。<br>
答：
延用上课使用的环境，表T 中的OBJECT_ID 中建有索引，现在使用两种方法进行查询：
1.1.1 使用索引查询
SQL> explain plan for select * from t where object_id=100;
Explained
SQL> select * from table(dbms_xplan.display(null,null,'typical'));
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 514881935
--------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Ti
--------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 63 | 6111 | 67 (0)| 00
| 1 | TABLE ACCESS BY INDEX ROWID| T | 63 | 6111 | 67 (0)| 00
|* 2 | INDEX RANGE SCAN | IDX_T_ID | 63 | | 3 (0)| 00
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("OBJECT_ID"=100)
14 rows selected
1.1.2 使用HINT 指定使用全表搜索查询
SQL> explain plan for select /*+FULL(T)*/ * from t where object_id=100;
Explained
SQL> select * from table(dbms_xplan.display(null,null,'typical'));
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 1601196873
--------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 63 | 6111 | 28636 (1)| 00:05:44 |
|* 1 | TABLE ACCESS FULL| T | 63 | 6111 | 28636 (1)| 00:05:44 |
--------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("OBJECT_ID"=100)
13 rows selected
从以上执行计划的成本 可以看到，索引搜索是全表搜索的1/400以上；
1.2 再看一个使用索引的性能比全表搜索差的演示：
建立一个索引
create index idx_t_status on t(status);
1.2.1 查询表中object_id<=100 （基本上是全表数据一起查询了）
1.2.2 按索引查询
SQL> explain plan for select /*+ index(t idx_t_status)*/ * from T WHERE status ='VALID';
Explained
SQL> select * from table(dbms_xplan.display(null,null,'typical'));
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 709710412
--------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)
--------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 2432K| 224M| 40808 (1)
| 1 | TABLE ACCESS BY INDEX ROWID| T | 2432K| 224M| 40808 (1)
|* 2 | INDEX RANGE SCAN | IDX_T_STATUS | 2432K| | 5792 (1)
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("STATUS"='VALID')
14 rows selected
1.2.3 没有指定使用索引，系统自动进行了全表搜索
SQL> explain plan for select * from T WHERE status ='VALID';
Explained
SQL> select * from table(dbms_xplan.display(null,null,'typical'));
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 1601196873
--------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 2432K| 224M| 28676 (1)| 00:05:45 |
|* 1 | TABLE ACCESS FULL| T | 2432K| 224M| 28676 (1)| 00:05:45 |
--------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("STATUS"='VALID')
13 rows selected
SQL>
从上面演示可以看出，没有使用索引的执行计划反而比使用了索引更优。是因为status ='VALID'
的选择性并不高，基本上所有数据都满足此条件。
****************************************************************************************************
2.分别给出一个Bitmap索引针对b-tree索引性能高和低的例子。<br>
再延用上面的数据示例，使用索引查询所有状态为status ='VALID' 的总数。
看到这时的执行成本 为： 5792
SQL> explain plan for select /*+ index(t idx_t_status)*/ count(0) from T WHERE status ='VALID';
Explained
SQL> select * from table(dbms_xplan.display(null,null,'typical'));
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 1684240785
--------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time
--------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 7 | 5792 (1)| 00:01:10
| 1 | SORT AGGREGATE | | 1 | 7 | |
|* 2 | INDEX RANGE SCAN| IDX_T_STATUS | 2432K| 16M| 5792 (1)| 00:01:10
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("STATUS"='VALID')
14 rows selected
SQL>
在表T 中，删除上一个索引，另外建立一个bitmap索引：
create Bitmap index idx_t_status on t(status);
SQL> explain plan for select count(0) from T where status ='VALID';
Explained
SQL> select * from table(dbms_xplan.display(null,null,'typical'));
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 1122805388
--------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)
--------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 7 | 61 (0)
| 1 | SORT AGGREGATE | | 1 | 7 |
| 2 | BITMAP CONVERSION COUNT | | 2432K| 16M| 61 (0)
|* 3 | BITMAP INDEX SINGLE VALUE| IDX_T_STATUS | | |
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
3 - access("STATUS"='VALID')
15 rows selected
SQL>
这时，系统自动走了BITMAP 索引，这时的执行成本 只有 61.优势明显。
****************************************************************************************************
3.演示DML操作导致位图索引锁定示例。<br>
3.1 建立示例数据
create table t2 as
SELECT OBJECT_ID FROM T
WHERE OBJECT_ID<10
insert into t2
SELECT OBJECT_ID FROM T
WHERE OBJECT_ID<10
select object_id,count(0) from t2
group by object_id
OBJECT_ID COUNT(0)
---------------------------
2 960
3 960
4 960
5 960
6 960
7 960
8 960
9 960
3.2建立bitmap 索引
create bitmap index idx_t2_id on t2(object_id);
3.3 测试BITMAP 锁机制
3.3.1 在一个窗口中修改数据，把 object_id in (2,3,4) 修改为1
SQL> select distinct sid from v$mystat s;
SID
----------
92
SQL> update t2 set object_id=1 where object_id in (2,3,4);
2880 rows updated
SQL>
3.3.2 再开3个窗口，插入OBJECT_ID 为2，3，4 的数据，看到，都已处在锁状态
SQL> select distinct sid from v$mystat s;
SID
----------
24
SQL> insert into t2 values(2);
SQL> select distinct sid from v$mystat s;
SID
----------
103
SQL> insert into t2 values(3);
SQL> select distinct sid from v$mystat s;
SID
----------
146
SQL> insert into t2 values(4);
再开一个窗口，把object_id =5 的也修改成1，这时看到，也是阻塞状态
SQL> select distinct sid from v$mystat s;
SID
----------
157
SQL> update t2 set object_id=1 where object_id =5;
我们再查询表：V$LOCK;
SELECT * FROM V$LOCK L WHERE L.TYPE IN ('TX','TM');
ADDR KADDR SID TYPE ID1 ID2 LMODE REQUEST CTIME BLOCK
1 00000002B13B6190 00000002B13B61E8 146 TX 655364 198442 0 4 430 0
2 00000002B13B66E8 00000002B13B6740 103 TX 655364 198442 0 4 448 0
3 00000002B13BD508 00000002B13BD560 157 TX 655364 198442 0 4 254 0
4 00000002B13BD968 00000002B13BD9C0 24 TX 655364 198442 0 4 455 0
5 0000000015FC4270 0000000015FC42D0 157 TM 94186 0 3 0 254 0
6 0000000015FC4270 0000000015FC42D0 146 TM 94186 0 3 0 430 0
7 0000000015FC4270 0000000015FC42D0 24 TM 91329 0 3 0 1437 0
8 0000000015FC4270 0000000015FC42D0 24 TM 94186 0 3 0 455 0
9 0000000015FC4270 0000000015FC42D0 103 TM 94186 0 3 0 448 0
10 0000000015FC4270 0000000015FC42D0 92 TM 94186 0 3 0 479 0
11 00000002A9EB3090 00000002A9EB3108 157 TX 458765 11761 6 0 254 0
12 00000002A8D0A528 00000002A8D0A5A0 146 TX 589855 74744 6 0 430 0
13 00000002A9EED250 00000002A9EED2C8 24 TX 65539 10414 6 0 1437 0
14 00000002A9EF1A28 00000002A9EF1AA0 103 TX 262165 10701 6 0 448 0
15 00000002A8D15D18 00000002A8D15D90 92 TX 655364 198442 6 0 479 1
可以看到第15条记录中，进程92有一个锁，
1-4记录中的 146，103，157，24 进程，都在等待申请中。
这就证明，位图索引在修改中，会把相同的索引段都加锁，而其它的相同索引段的对应操作（添加，修改）都会被阻塞。
****************************************************************************************************
4.创建一个全文索引，比较它和传统的模糊查询的性能。<br>
4.1 测试环境
select count(0) from company
union all
select count(0) from company_text_idx
COUNT(0)
-----------
9999
9999
4.2 在传统表中建立索引
create index idx_name on company(name);
4.3 在全文索引测试表中建立全文索引
BEGIN
ctx_ddl.create_preference ('my_lexer', 'chinese_vgram_lexer');
END;
/
CREATE INDEX idx_comp_text_idx ON company_text_idx(name) indextype is ctxsys.context parameters('lexer my_lexer');
4.4全文搜索执行计划：
从执行计划中可以看到，查询9条记录，成本 为：6；字节为 79130k，唯一性读为：473
SQL> select * from company_text_idx where contains(name,'王飞')>0;
已选择9行。
执行计划
----------------------------------------------------------
Plan hash value: 3789291194
---------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 5 | 79130 | 6 (0)| 00:00:01 |
| 1 | TABLE ACCESS BY INDEX ROWID| COMPANY_TEXT_IDX | 5 | 79130 | 6 (0)| 00:00:01 |
|* 2 | DOMAIN INDEX | IDX_COMP_TEXT_IDX | | | 4 (0)| 00:00:01 |
---------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("CTXSYS"."CONTAINS"("NAME",'王飞')>0)
Note
-----
- dynamic sampling used for this statement (level=2)
统计信息
----------------------------------------------------------
137 recursive calls
0 db block gets
473 consistent gets
0 physical reads
0 redo size
34324 bytes sent via SQL*Net to client
22618 bytes received via SQL*Net from client
78 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
9 rows processed
4.5传统计索引搜索计划：
执行成本 为9，唯一性读为91，字节为 138K,
成本比全文搜索差一些。但读取的字节数少，唯一性读也少。从这里也可以看到，全文检索使用了更大的存储空间。
SQL> select * from company where name like '王飞%';
已选择9行。
执行计划
----------------------------------------------------------
Plan hash value: 1068597349
------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 9 | 138K| 9 (0)| 00:00:01 |
| 1 | TABLE ACCESS BY INDEX ROWID| COMPANY | 9 | 138K| 9 (0)| 00:00:01 |
|* 2 | INDEX RANGE SCAN | IDX_NAME | 9 | | 2 (0)| 00:00:01 |
------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("NAME" LIKE '王飞%')
filter("NAME" LIKE '王飞%')
Note
-----
- dynamic sampling used for this statement (level=2)
统计信息
----------------------------------------------------------
9 recursive calls
0 db block gets
91 consistent gets
0 physical reads
0 redo size
34236 bytes sent via SQL*Net to client
22618 bytes received via SQL*Net from client
78 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
9 rows processed
SQL>
4.6 再换一个传统查询中，包含条件的执行计划。
这时可以看到，无法使用索引了，成本达到了 254,唯一性读也达到了987。
SQL> select * from company where name like '%王飞%';
已选择9行。
执行计划
----------------------------------------------------------
Plan hash value: 1376141687
--------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 26 | 401K| 254 (0)| 00:00:04 |
|* 1 | TABLE ACCESS FULL| COMPANY | 26 | 401K| 254 (0)| 00:00:04 |
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter("NAME" IS NOT NULL AND "NAME" LIKE '%王飞%')
Note
-----
- dynamic sampling used for this statement (level=2)
统计信息
----------------------------------------------------------
5 recursive calls
0 db block gets
987 consistent gets
0 physical reads
0 redo size
34324 bytes sent via SQL*Net to client
22618 bytes received via SQL*Net from client
78 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
9 rows processed
SQL>
从上面的3个执行计划中可以看出，全文检索的性能及搜索灵活性在包含条件下的查询，是更高的。
但也使用了更多的存储空间。
****************************************************************************************************
5.分别演示分区索引的性能优化全局索引和差于全局索引的示例，并分析原因。<br>（边城日志 2013/11/15 15：50）
5.1构建演示环境：
在一现有示例表中，基本每月有3000条记录，现另建立表个表，按季度分区保存。
select to_char(po_dt,'yyyymm') as ym,count(0) from po
group by to_char(po_dt,'yyyymm')
order by
to_char(po_dt,'yyyymm')
YM COUNT(0)
----------------
1 201001 3099
2 201002 2800
3 201003 3100
4 201004 3000
5 201005 3100
6 201006 3000
7 201007 3100
8 201008 3100
9 201009 3000
10 201010 3100
11 201011 3000
12 201012 3100
13 201101 3100
14 201102 2800
15 201103 3100
16 201104 3000
17 201105 1500
--3个月一个分区
CREATE TABLE PO_R
( POID NUMBER(10,0),
QTY NUMBER(10,2),
AMOUNT NUMBER(10,2),
AUDIT_FLG NUMBER(1,0) DEFAULT 0,
PO_DT DATE
)
PARTITION BY RANGE (PO_DT)
INTERVAL (NUMTOYMINTERVAL(3, 'MONTH'))
(PARTITION P1 VALUES LESS THAN (TO_DATE('2010-01-01', 'YYYY-MM-DD')))
;
insert into po_r
select * from PO
commit;
每个分区的数据量
select t.partition_name,t.num_rows
from all_tab_partitions t where table_name='PO_R'
PARTITION_NAME NUM_ROWS
-------------------------------
1 P1 0
2 SYS_P141 8999
3 SYS_P142 9100
4 SYS_P143 9200
5 SYS_P144 9200
6 SYS_P145 9000
7 SYS_P146 4500
5.2.我用两种索引来进行对比：
5.2.1 全局索引
create index IDX_PO_R_ID_GLOBAL on PO_R (poid);
5.2.2 分区索引
create index IDX_PO_R_ID_LOCAL on PO_R (POID)
local;
以下为两次不同索引下的执行计划：
全局索引：成本：1869，唯一性读2403，总字节：187K
分区索引：成本：1871，唯一性读2412，总字节：187K
对比看出，全局索引稍胜于分区索引
SQL> set autotrace on
SQL> set autotrace traceonly
SQL> select /*+INDEX(p IDX_PO_R_ID_GLOBAL)*/ * from po_r p where poid between 500 and 7900;
7401 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 1901754288
-------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Pstart| Pstop |
-------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 7402 | 187K| 1868 (1)| 00:00:23 | | |
| 1 | TABLE ACCESS BY GLOBAL INDEX ROWID| PO_R | 7402 | 187K| 1868 (1)| 00:00:23 | ROWID | ROWID |
|* 2 | INDEX RANGE SCAN | IDX_PO_R_ID_GLOBAL | 7402 | | 21 (0)| 00:00:01 | | |
-------------------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("POID">=500 AND "POID"<=7900)
Statistics
----------------------------------------------------------
1 recursive calls
0 db block gets
2403 consistent gets
19 physical reads
0 redo size
231787 bytes sent via SQL*Net to client
5942 bytes received via SQL*Net from client
495 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
7401 rows processed
SQL> select /*+INDEX(p IDX_PO_R_ID_LOCAL)*/ * from po_r p where poid between 500 and 7900;
7401 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 3357987540
------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Pstart| Pstop |
------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 7402 | 187K| 1871 (1)| 00:00:23 | | |
| 1 | PARTITION RANGE ALL | | 7402 | 187K| 1871 (1)| 00:00:23 | 1 |1048575|
| 2 | TABLE ACCESS BY LOCAL INDEX ROWID| PO_R | 7402 | 187K| 1871 (1)| 00:00:23 | 1 |1048575|
|* 3 | INDEX RANGE SCAN | IDX_PO_R_ID_LOCAL | 7402 | | 24 (0)| 00:00:01 | 1 |1048575|
------------------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
3 - access("POID">=500 AND "POID"<=7900)
Statistics
----------------------------------------------------------
1 recursive calls
0 db block gets
2412 consistent gets
21 physical reads
0 redo size
231787 bytes sent via SQL*Net to client
5942 bytes received via SQL*Net from client
495 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
7401 rows processed
SQL>
5.3再看一个演示，也是全局索引稍胜于分区索引
SQL> select /*+INDEX(p IDX_PO_R_ID_LOCAL)*/ * from po_r p where poid = 80;
Execution Plan
----------------------------------------------------------
Plan hash value: 3357987540
------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Pstart| Pstop |
------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 26 | 9 (0)| 00:00:01 | | |
| 1 | PARTITION RANGE ALL | | 1 | 26 | 9 (0)| 00:00:01 | 1 |1048575|
| 2 | TABLE ACCESS BY LOCAL INDEX ROWID| PO_R | 1 | 26 | 9 (0)| 00:00:01 | 1 |1048575|
|* 3 | INDEX RANGE SCAN | IDX_PO_R_ID_LOCAL | 1 | | 8 (0)| 00:00:01 | 1 |1048575|
------------------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
3 - access("POID"=80)
Statistics
----------------------------------------------------------
1 recursive calls
0 db block gets
14 consistent gets
1 physical reads
0 redo size
811 bytes sent via SQL*Net to client
519 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
1 rows processed
SQL> select /*+INDEX(p IDX_PO_R_ID_GLOBAL)*/ * from po_r p where poid = 80;
Execution Plan
----------------------------------------------------------
Plan hash value: 1901754288
-------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Pstart| Pstop |
-------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 1 | 26 | 2 (0)| 00:00:01 | | |
| 1 | TABLE ACCESS BY GLOBAL INDEX ROWID| PO_R | 1 | 26 | 2 (0)| 00:00:01 | ROWID | ROWID |
|* 2 | INDEX RANGE SCAN | IDX_PO_R_ID_GLOBAL | 1 | | 1 (0)| 00:00:01 | | |
-------------------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("POID"=80)
Statistics
----------------------------------------------------------
1 recursive calls
0 db block gets
4 consistent gets
1 physical reads
0 redo size
811 bytes sent via SQL*Net to client
519 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
1 rows processed
SQL>
5.4是不是全局索引就一定比分区索引更优呢，下面 我使用PO_DT 字段来建立索引。
从下面的执行计划中可以看出：
这时的分区索引是稍优全局索引的。
所以就性能上来说，并不能武断的说明哪咱索引更优。因为不管是哪种索引，在查询时，都是查询某一段(或一条)。
然后查询对应一段(或一条)的数据
SQL> select * from po_r p where PO_DT=TO_DATE('20100606','YYYYMMDD');
100 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 3172462535
-----------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Pstart| Pstop |
-----------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 100 | 2600 | 4 (0)| 00:00:01 | | |
| 1 | PARTITION RANGE SINGLE | | 100 | 2600 | 4 (0)| 00:00:01 | 3 | 3 |
| 2 | TABLE ACCESS BY LOCAL INDEX ROWID| PO_R | 100 | 2600 | 4 (0)| 00:00:01 | 3 | 3 |
|* 3 | INDEX RANGE SCAN | IDX_POR_DT_LOCAL | 100 | | 1 (0)| 00:00:01 | 3 | 3 |
-----------------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
3 - access("PO_DT"=TO_DATE(' 2010-06-06 00:00:00', 'syyyy-mm-dd hh24:mi:ss'))
Statistics
----------------------------------------------------------
1 recursive calls
0 db block gets
18 consistent gets
1 physical reads
0 redo size
4026 bytes sent via SQL*Net to client
585 bytes received via SQL*Net from client
8 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
100 rows processed
SQL> select * from po_r p where PO_DT=TO_DATE('20100606','YYYYMMDD');
100 rows selected.
Execution Plan
----------------------------------------------------------
Plan hash value: 3915168380
------------------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Pstart| Pstop |
------------------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 100 | 2600 | 5 (0)| 00:00:01 | | |
| 1 | TABLE ACCESS BY GLOBAL INDEX ROWID| PO_R | 100 | 2600 | 5 (0)| 00:00:01 | 3 | 3 |
|* 2 | INDEX RANGE SCAN | IDX_POR_DT_GLOBAL | 100 | | 1 (0)| 00:00:01 | | |
------------------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("PO_DT"=TO_DATE(' 2010-06-06 00:00:00', 'syyyy-mm-dd hh24:mi:ss'))
Statistics
----------------------------------------------------------
1 recursive calls
0 db block gets
18 consistent gets
2 physical reads
0 redo size
4026 bytes sent via SQL*Net to client
585 bytes received via SQL*Net from client
8 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
100 rows processed
SQL>
【性能优化】之 BITMAP 及分区表 的演示相关推荐
- Android 图片性能优化：Bitmap
一.引入 图片作为内存消耗大户,一直是开发人员尝试优化的重点对象.Bitmap的内存从3.0以前的位于native,到后来改成jvm,再到8.0又改回到native.jvm每个进程都有内存上限,而na ...
- SQL SERVER 性能优化四： 创建分区表
1.整体介绍 1.1 分区表概念:分区表值得是逻辑上是一个表,物理上被存储到不同的磁盘文件中. 1.2 优势:提高查询性能:提高稳定性:便于管理:对于大数据量表备份更方便. 1.3 建立分区表主要包含 ...
- Android性能优化：那些关于Bitmap图片资源优化的小事
前言 在 Android开发中,性能优化策略十分重要 本文主要讲解性能优化中的Bitmap 使用优化,希望你们会喜欢 Carson带你学Android性能优化系列文章: Android性能优化:性能优 ...
- android 性能优化---（5）Bitmap图片资源优化
前言 在 Android开发中,性能优化策略十分重要 本文主要讲解性能优化中的Bitmap 使用优化,希望你们会喜欢 目录 1. 优化原因 即 为什么要优化图片Bitmap资源,具体如下图: 2. ...
- Android性能优化系列之Bitmap图片优化
在Android开发过程中,Bitmap往往会给开发者带来一些困扰,因为对Bitmap操作不慎,就容易造成OOM(Java.lang.OutofMemoryError - 内存溢出),本篇博客,我们将 ...
- MySql性能优化之分区表
分区表是一个独立的逻辑表,但是底层是由多个物理子表组成,MySQL实现分区表的方式,是对底层表的封装,意味着索引也是按照分区的子表定义的,而没有全局索引.所以分区表可以看成合并表的升级,是做了性能优化 ...
- Android性能优化系列：Bitmap
文章目录 Bitmap 简介 Bitmap 的创建 不同系统版本 Bitmap 的内存分配策略 Bitmap 内存占用计算 在电脑查看的图片大小和运行内存大小区别 图片占用内存计算 Bitmap 内存 ...
- Oracle数据库性能优化艺术（第五期） 第7周 索引和分区（包括11g下新的组合分区）
1.分别给出一个B-tree索引针对全表扫描性能高和低的例子. B-tree比FTS性能高的例子: SQL> drop table t purge; Table dropped. SQL> ...
- Oracle服务器性能优化
几个简单的步骤大幅提高Oracle性能--我优化数据库的三板斧 数据库优化的讨论可以说是一个永恒的主题.资深的Oracle优化人员通常会要求提出性能问题的人对数据库做一个statspack,贴出数据库 ...
最新文章
- 【组队学习】【34期】零基础学python编程思维
- android .so文件详解以及兼容性
- 福利满满 | 天元MegEngine贡献者计划全面启动！
- web页面优化之动态加载js和文件
- 在web应用程序中使用MemcachedClient
- iPhone 12 Pro火爆程度超预期 苹果紧急向关键组件厂商加单
- PostgreSQL修炼之道：从小工到专家. 3.1 SQL语句语法简介
- 哈佛大学幸福课笔记一
- python解析FreeMind思维导图
- android 9.0 c7Pro,透心凉！三星Galaxy C7 Pro上线，还内置热管
- 拼多多数据分析一二三面面经（HR面后综排挂）
- Rxjava +Retrofit 你需要掌握的几个技巧，Retrofit缓存，RxJava封装，统一对有无网络处理,异常处理， 返回结果问题
- 错过了愚人节，还有清明节
- 熊孩子乱敲键盘攻破linux桌面,“熊孩子”乱敲键盘攻破了Linux桌面 大神：17年前我就警告过...
- Servlet中关于Session数据存储遇到的数据转换问题
- ChIPseeker入门到精通
- 乡村振兴--交通建设
- 最大似然估计(MLE)与最小二乘估计(LSE)的区别
- Anaconda安装教程傻瓜教程
- mtkwin10驱动_修复：Win10系统MTK（MediaTek）VCOM USB驱动程序错误