B树索引学习总结。

   题记：今天是2014-01-13，在春节之前打算把oracle索引技术深入研究一下，在此记录一下学习笔记。今天我学习的内容是B树索引知识。

                       B树索引深入总结。

 首先看一下网络上一个B树索引图，如下：

 1、oracle如何使用B树索引？

B树索引有根节点块，分支节点块，叶子节点块组成。一般情况下B树索引是oracle默认创建索引使用的类型。使用B树索引可以提高数据的检索速度，可以强制执行主键和唯一键的约束性，对于减少通过主键和外键约束关联的父表和字表锁的问题。

 当检索数据列位于索引指定列的时候，只需要访问索引块即可完成数据的展现，这非常提高数据的检索速率；

 当检索数据列不完全位于索引指定列的时候，将会访问索引块以及数据块进而将数据进行展现；

 最后一种是不适用索引，而是通过检索表块完成数据的检索；

下面分别对这三种情况进行试验说明：

NO.1：对于该情况所有字段都位于索引 段中，一般设计到index range scan表示要扫B树索引的叶子子节点一段数据；另一个是index fast full scan，表示要扫描所有索引的叶子子节点数据，一般用在count中如下：

SQL> create table emp(
  2  empno number(4),
  3  ename varchar2(10),
  4  job varchar2(9),
  5  mgr number(4),
  6  hiredate date,
  7  sal number(7,2),
  8  comm number(7,2),
  9  deptno number(2)
 10  );

Table created.
SQL> create index emp_pk on emp(empno);

Index created.

SQL> insert into emp select * from scott.emp;

14 rows created.

SQL> insert into emp select * from scott.emp;

14 rows created.

SQL> set autotrace trace 
SQL> select empno from emp where empno=7369;


Execution Plan
----------------------------------------------------------
Plan hash value: 745440807

---------------------------------------------------------------------------
| Id  | Operation        | Name   | Rows  | Bytes | Cost (%CPU)| Time     |
---------------------------------------------------------------------------
|   0 | SELECT STATEMENT |        |     2 |    26 |     1   (0)| 00:00:01 |
|*  1 |  INDEX RANGE SCAN| EMP_PK |     2 |    26 |     1   (0)| 00:00:01 |
---------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   1 - access("EMPNO"=7369)

Note
-----
   - dynamic sampling used for this statement (level=2)


Statistics
----------------------------------------------------------
         22  recursive calls
          0  db block gets
         36  consistent gets
          0  physical reads
          0  redo size
        579  bytes sent via SQL*Net to client
        523  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          3  sorts (memory)
          0  sorts (disk)
          2  rows processed

SQL> 

 

有个问题，在这个操作过程中，对应一下上图，该过程会扫描几个数据块？嘿嘿，可以思考一下。

以上图为例，每个大方框都代表一个数据块，那么如果要获得0号数据信息，那么就会读0B1数据块、B1数据块、L1数据块，如果为了确保L1数据都要验证是否满足需求因此需要读整个L1块的数据，这就是index range scan；
注意：

db block gets=db block gets from cache+db block gets direct

consistent gets=consistent gets from cache+consistent get direct

physical reads=physical reads  cache+physical

在看一下index fast full scan：

 

SQL> select count(empno) from emp;


Execution Plan
----------------------------------------------------------
Plan hash value: 3381701820

---------------------------------------------------------------------------
| Id  | Operation        | Name   | Rows  | Bytes | Cost (%CPU)| Time     |
---------------------------------------------------------------------------
|   0 | SELECT STATEMENT |        |     1 |    13 |     1   (0)| 00:00:01 |
|   1 |  SORT AGGREGATE  |        |     1 |    13 |            |          |
|   2 |   INDEX FULL SCAN| EMP_PK |    28 |   364 |     1   (0)| 00:00:01 |
---------------------------------------------------------------------------

Note
-----
   - dynamic sampling used for this statement (level=2)


Statistics
----------------------------------------------------------
         18  recursive calls
          0  db block gets
         34  consistent gets
          0  physical reads
          0  redo size
        530  bytes sent via SQL*Net to client
        523  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          3  sorts (memory)
          0  sorts (disk)
          1  rows processed

no，貌似没有index fast full scan而是走了单块读的index full scan；

 

SQL> set autotrace trace
SQL> select /*+index_ffs(emp emp_pk)*/ count(empno) from emp;


Execution Plan
----------------------------------------------------------
Plan hash value: 3578136827

------------------------------------------------------------------------
| Id  | Operation             | Name   | Rows  | Cost (%CPU)| Time     |
------------------------------------------------------------------------
|   0 | SELECT STATEMENT      |        |     1 |     2   (0)| 00:00:01 |
|   1 |  SORT AGGREGATE       |        |     1 |            |          |
|   2 |   INDEX FAST FULL SCAN| EMP_PK |    14 |     2   (0)| 00:00:01 |
------------------------------------------------------------------------

Note
-----
   - dynamic sampling used for this statement (level=2)


Statistics
----------------------------------------------------------
          4  recursive calls
          0  db block gets
         23  consistent gets
          0  physical reads
          0  redo size
        530  bytes sent via SQL*Net to client
        523  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          0  sorts (memory)
          0  sorts (disk)
          1  rows processed

 NO.2:部分列不存在索引列中。

SQL> select empno,ename,job from emp where empno=7369;


Execution Plan
----------------------------------------------------------
Plan hash value: 122628344

--------------------------------------------------------------------------------------
| Id  | Operation                   | Name   | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |        |     2 |    52 |     1   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| EMP    |     2 |    52 |     1   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN          | EMP_PK |     2 |       |     1   (0)| 00:00:01 |
--------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   2 - access("EMPNO"=7369)

Note
-----
   - dynamic sampling used for this statement (level=2)


Statistics
----------------------------------------------------------
          7  recursive calls
          0  db block gets
         26  consistent gets
          0  physical reads
          0  redo size
        735  bytes sent via SQL*Net to client
        523  bytes received via SQL*Net from client
          2  SQL*Net roundtrips to/from client
          0  sorts (memory)
          0  sorts (disk)
          2  rows processed

SQL> 

可以看到当查询列不存在于索引列的时候，将会通过rowid来对数据表进行访问数据，因为B数索引就是通过rowid和建立索引的列值组合的。注意当该表中存在多条记录的时候consistent gets并不能代表真正的数据块的一直性读，因为重新读取的块被收集在缓存区被钉住计数（buffer is pinned count）。

对于rowid的理解请参考我的另一篇笔记：

http://blog.csdn.net/rhys_oracle/article/details/11715735

NO。3：不适用索引的情况。

SQL> select * from emp;

28 rows selected.


Execution Plan
----------------------------------------------------------
Plan hash value: 3956160932

--------------------------------------------------------------------------
| Id  | Operation         | Name | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------
|   0 | SELECT STATEMENT  |      |    28 |  2436 |     6   (0)| 00:00:01 |
|   1 |  TABLE ACCESS FULL| EMP  |    28 |  2436 |     6   (0)| 00:00:01 |
--------------------------------------------------------------------------

Note
-----
   - dynamic sampling used for this statement (level=2)


Statistics
----------------------------------------------------------
         17  recursive calls
          0  db block gets
         54  consistent gets
          0  physical reads
          0  redo size
       2411  bytes sent via SQL*Net to client
        534  bytes received via SQL*Net from client
          3  SQL*Net roundtrips to/from client
          3  sorts (memory)
          0  sorts (disk)
         28  rows processed

这就是oracle cbo采用全表扫描方式。注意：有时候采用全表扫描可能比采用索引效率更高。

2、准备创建B树索引：

命名规则：

主键索引名应该包含表名和一个后缀，如：_PK；

唯一索引应该包含表名和一个后缀，如：_UKN,N代表一个数字；

外键索引应该包含外键表和一个后缀，如_FKN，N代表一个数字；

对于不适用约束的索引，适用表名和一个后缀，如_IDXN,N代表一个数字；

基于函数的索引的名称应该适用一个表名和一个后缀，如_FCN,N代表一个数字；

只所以采用这种命名方法是便于查看和便于问题分析。

估算索引大小；

因为当在一个大表上创建索引或是在一个业务量增长非常迅速的表上维护索引时，该索引同样需要消耗资源，当表空间资源不足的时候，就有可能导致数据库挂起。

对于估算索引一般采用如下步骤：

1、收集最新的表的统计信息

2、使用dbms_space.create_index_cost进行 索引估算。下面测试如

估算需要创建索引的大小，如下：

 

SQL> EXEC DBMS_STATS.gather_TABLE_STATS(OWNNAME=>'AMY',TABNAME=>'T',CASCADE=>TRUE);

PL/SQL procedure successfully completed.

SQL> select index_name,table_name from user_indexes where table_name='T';

no rows selected

SQL> SET SERVEROUT ON   
SQL> VAR USED_BYTES NUMBER
SQL> VAR ALLOC_BYTES NUMBER
SQL> EXEC DBMS_SPACE.CREATE_INDEX_COST('CREATE INDEX T_IDX1 ON T(A)',:USED_BYTES,:ALLOC_BYTES);

PL/SQL procedure successfully completed.

SQL> PRINT USED_BYTES

USED_BYTES
----------
   1135000

SQL> PRINT ALLOC_BYTES

ALLOC_BYTES
-----------
    6291456

SQL> CREATE INDEX T_IDX1 ON T(A);

Index created.


SQL> SELECT SEGMENT_NAME,BYTES FROM USER_SEGMENTS WHERE SEGMENT_NAME='T_IDX1';

SEGMENT_NAME              BYTES
-------------------- ----------
T_IDX1                  4423680

SQL> 

索引与表数据分离，单独创建表空间，这样可以单独维护索引如备份和恢复，但是减少i/0其实更应该放入不同的磁盘上面；

SQL> create tablespace reporting_index datafile '+DATAGROUP1' size 512M autoextend on next 1M maxsize 1G extent management local uniform size 128K segment space management auto nologging;

Tablespace created.
SQL> DROP INDEX T_IDX1;

Index dropped.

SQL> CREATE INDEX T_IDX1 ON T(A) TABLESPACE REPORTING_INDEX;

Index created.

SQL>

SQL> alter table emp add constraint emp_pk primary key(empno) using index tablespace reporting_index;

Table altered.

SQL> select index_name,table_name from user_indexes where table_name='EMP';

INDEX_NAME                     TABLE_NAME
------------------------------ ------------------------------
EMP_PK                         EMP

SQL> alter table emp add constraint emp_uk unique(empno,ename) using index tablespace reporting_index;

Table altered.
SQL> DROP INDEX EMP_IDX2;

Index dropped.

SQL> alter table dept add constraint dept_pk primary key(deptno) using index tablespace reporting_index;

Table altered.

SQL> CREATE INDEX EMP_IDX2 ON EMP(DEPTNO) TABLESPACE REPORTING_INDEX;

Index created.

SQL> select index_name,INDEX_TYPE,table_name,tablespace_name,status from user_indexes where table_name in('T','EMP','DEPT');

INDEX_NAME                     INDEX_TYPE                  TABLE_NAME                     TABLESPACE_NAME                STATUS
------------------------------ --------------------------- ------------------------------ ------------------------------ --------
T_IDX1                         NORMAL                      T                              REPORTING_INDEX                VALID
EMP_PK                         NORMAL                      EMP                            REPORTING_INDEX                VALID
EMP_UK                         NORMAL                      EMP                            REPORTING_INDEX                VALID
EMP_IDX2                       NORMAL                      EMP                            REPORTING_INDEX                VALID
DEPT_PK                        NORMAL                      DEPT                           REPORTING_INDEX                VALID

SQL> col column_name for a20
SQL> col column_position for 9999
SQL> r
  1* SELECT INDEX_NAME,COLUMN_NAME,COLUMN_POSITION FROM USER_ind_columns where table_name in ('T','EMP','DEPT') ORDER by index_name,column_position

INDEX_NAME                     COLUMN_NAME          COLUMN_POSITION
------------------------------ -------------------- ---------------
DEPT_PK                        DEPTNO                             1
EMP_IDX2                       DEPTNO                             1
EMP_PK                         EMPNO                              1
EMP_UK                         EMPNO                              1
EMP_UK                         ENAME                              2
T_IDX1                         A                                  1

6 rows selected.

SQL> 
SQL> 

修改索引状态：

SQL> alter index emp_idx2 invisible;

Index altered.

SQL> select index_name,status,visibility from user_indexes;

INDEX_NAME                     STATUS   VISIBILIT
------------------------------ -------- ---------
EMP_UK                         VALID    VISIBLE
EMP_PK                         VALID    VISIBLE
EMP_IDX2                       VALID    INVISIBLE
DEPT_PK                        VALID    VISIBLE
T_IDX1                         VALID    VISIBLE

SQL> alter index emp_idx2 visible;

Index altered.

SQL> select index_name,status,visibility from user_indexes;

INDEX_NAME                     STATUS   VISIBILIT
------------------------------ -------- ---------
EMP_UK                         VALID    VISIBLE
EMP_PK                         VALID    VISIBLE
EMP_IDX2                       VALID    VISIBLE
DEPT_PK                        VALID    VISIBLE
T_IDX1                         VALID    VISIBLE

SQL> 
SQL> alter index emp_idx2 unusable;

Index altered.

SQL> select index_name,status,visibility from user_indexes;

INDEX_NAME                     STATUS   VISIBILIT
------------------------------ -------- ---------
EMP_UK                         VALID    VISIBLE
EMP_PK                         VALID    VISIBLE
EMP_IDX2                       UNUSABLE VISIBLE
DEPT_PK                        VALID    VISIBLE
T_IDX1                         VALID    VISIBLE

SQL> alter index emp_idx2 rebuild;

Index altered.

SQL> select index_name,status,visibility from user_indexes;

INDEX_NAME                     STATUS   VISIBILIT
------------------------------ -------- ---------
EMP_UK                         VALID    VISIBLE
EMP_PK                         VALID    VISIBLE
EMP_IDX2                       VALID    VISIBLE
DEPT_PK                        VALID    VISIBLE
T_IDX1                         VALID    VISIBLE

SQL> 
SQL> 
SQL> 

 

判断外键索引是否存在：

方法一：

SET VERIFY OFF;
SET LINESIZE 200;
COL IND_COLUMN FOR A25;
COL CONS_COLUMN FOR A25;
COL TAB_NAME FOR A20;
COL OWNER FOR A25;
select distinct a.owner owner,
                a.constraint_name cons_name,
                a.table_name tab_name,
                b.column_name cons_column,
                nvl(c.column_name, '***Check index****') ind_column,
                c.index_name
  from dba_constraints a, dba_cons_columns b, dba_ind_columns c
 where constraint_type = 'R'
   and a.owner = upper('&&user_name')
   and a.owner = b.owner
   and a.constraint_name = b.constraint_name
   and b.column_name = c.column_name(+)
   and b.table_name = c.table_name(+)
   and b.position = c.column_position(+)
 order by tab_name, ind_column;


#############################

SQL> conn amy/rhys
Connected.
SQL> drop index emp_idx2;

Index dropped.

SQL> conn sys/root as sysdba
Connected.
SQL> SET VERIFY OFF;
SET LINESIZE 200;
SQL> SQL> COL IND_COLUMN FOR A25;
SQL> COL CONS_COLUMN FOR A25;
SQL> COL TAB_NAME FOR A20;
SQL> COL OWNER FOR A25;
SQL> select distinct a.owner owner,
  2                  a.constraint_name cons_name,
  3                  a.table_name tab_name,
                b.column_name cons_column,
  4    5                  nvl(c.column_name, '***Check index****') ind_column,
  6                  c.index_name
  7    from dba_constraints a, dba_cons_columns b, dba_ind_columns c
  8   where constraint_type = 'R'
  9     and a.owner = upper('&&user_name')
 10     and a.owner = b.owner
 11     and a.constraint_name = b.constraint_name
 12     and b.column_name = c.column_name(+)
 13     and b.table_name = c.table_name(+)
 14     and b.position = c.column_position(+)
 15   order by tab_name, ind_column;


OWNER                     CONS_NAME                      TAB_NAME             CONS_COLUMN               IND_COLUMN                INDEX_NAME
------------------------- ------------------------------ -------------------- ------------------------- ------------------------- ------------------------------
AMY                       EMP_FK                         EMP                  DEPTNO                    ***Check index****

SQL> SQL> 

 

 

方法一只是判断用户，对于单列约束。如下判断多列约束，另外还可以检测外键上的位图索引的存在，位图索引不能防止锁的发生；

####################3

col cols for a30;
col index_type for a10;
SET VERIFY OFF;
SET LINESIZE 200;
COL IND_COLUMN FOR A25;
COL CONS_COLUMN FOR A25;
COL index_NAME FOR A20;
COL OWNER FOR A25;
select case
         when ind.index_name is not null then
          case
            when ind.index_type in ('BITMAP') then
             '**Bitmp idx**'
            else
             'indexed'
          end
         else
          '**Check idx**'
       end checker,
       ind.index_type,
       cons.owner,
       cons.table_name,
       ind.index_name,
       cons.constraint_name,
       cons.cols
  from (select c.owner,
               c.table_name,
               c.constraint_name,
               listagg(cc.column_name, ',') within group(order by cc.column_name) cols
          from dba_constraints c, dba_cons_columns cc
         where c.owner = cc.owner
           and c.owner = upper('&&schema')
           and c.constraint_name = cc.constraint_name
           and c.constraint_type = 'R'
         group by c.owner, c.table_name, c.constraint_name) cons
  left outer join (select table_owner,
                          table_name,
                          index_name,
                          index_type,
                          cbr,
                          listagg(column_name, ',') within group(order by column_name) cols
                     from (select ic.table_owner,
                                  ic.table_name,
                                  ic.index_name,
                                  ic.column_name,
                                  ic.column_position,
                                  i.index_type,
                                  connect_by_root(ic.column_name) cbr
                             from dba_ind_columns ic, dba_indexes i
                            where ic.table_owner = upper('&&schema')
                              and ic.table_owner = i.table_owner
                              and ic.table_name = i.table_name
                              and ic.index_name = i.index_name
                           connect by prior ic.column_position - 1 =
                                       ic.column_position
                                  and prior ic.index_name = ic.index_name)
                    group by table_owner,
                             table_name,
                             index_name,
                             index_type,
                             cbr) ind
    on cons.cols = ind.cols
   and cons.table_name = ind.table_name
   and cons.owner = ind.table_owner
 order by checker, cons.owner, cons.table_name;


 

#############################

eg：

SQL> col cols for a30;
SQL> col index_type for a10;
SQL> SET VERIFY OFF;
SQL> SET LINESIZE 200;
COL IND_COLUMN FOR A25;
SQL> SQL> COL CONS_COLUMN FOR A25;
SQL> COL index_NAME FOR A20;
SQL> COL OWNER FOR A25;
SQL> select case
  2           when ind.index_name is not null then
  3            case
  4              when ind.index_type in ('BITMAP') then
  5               '**Bitmp idx**'
  6              else
             'indexed'
  7    8            end
  9           else
 10            '**Check idx**'
       end checker,
 11   12         ind.index_type,
 13         cons.owner,
 14         cons.table_name,
 15         ind.index_name,
 16         cons.constraint_name,
 17         cons.cols
 18    from (select c.owner,
 19                 c.table_name,
 20                 c.constraint_name,
 21                 listagg(cc.column_name, ',') within group(order by cc.column_name) cols
 22            from dba_constraints c, dba_cons_columns cc
         where c.owner = cc.owner
 23   24             and c.owner = upper('&&schema')
 25             and c.constraint_name = cc.constraint_name
 26             and c.constraint_type = 'R'
 27           group by c.owner, c.table_name, c.constraint_name) cons
 28    left outer join (select table_owner,
 29                            table_name,
 30                            index_name,
 31                            index_type,
 32                            cbr,
 33                            listagg(column_name, ',') within group(order by column_name) cols
 34                       from (select ic.table_owner,
 35                                    ic.table_name,
 36                                    ic.index_name,
 37                                    ic.column_name,
 38                                    ic.column_position,
 39                                    i.index_type,
 40                                    connect_by_root(ic.column_name) cbr
 41                               from dba_ind_columns ic, dba_indexes i
 42                              where ic.table_owner = upper('&&schema')
 43                                and ic.table_owner = i.table_owner
 44                                and ic.table_name = i.table_name
 45                                and ic.index_name = i.index_name
 46                             connect by prior ic.column_position - 1 =
 47                                         ic.column_position
 48                                    and prior ic.index_name = ic.index_name)
 49                      group by table_owner,
 50                               table_name,
 51                               index_name,
 52                               index_type,
 53                               cbr) ind
 54      on cons.cols = ind.cols
 55     and cons.table_name = ind.table_name
 56     and cons.owner = ind.table_owner
 57   order by checker, cons.owner, cons.table_name;


CHECKER       INDEX_TYPE OWNER                     TABLE_NAME                     INDEX_NAME           CONSTRAINT_NAME                COLS
------------- ---------- ------------------------- ------------------------------ -------------------- ------------------------------ ------------------------------
**Check idx**            AMY                       EMP                                                 EMP_FK                         DEPTNO

SQL> SQL>