oracle ora 47306,Oracle SQL提示含义与示例 --- 分布式查询和并行提示
《Oracle 高性能SQL引擎剖析：Oracle SQL 优化与调优技术详解》一书的附录部分。作为对该书的补充，帮助读者理解和掌握“提示”这一项在SQL优化中使用的这一重要辅助手段。
语法：DRIVING_SITE([])
描述：指示优化器选择那个数据库作为分布式查询中的驱动站点，即将语句放在该站点上执行。未指定参数时，采用本地数据库。
HELLODBA.COM>exec sql_explain('select /*+ driving_site(rt) */count(*) from t_tables lt, t_tables@ora11r2 rt where lt.owner = rt.owner and lt.table_name = rt.table_name', 'TYPICAL NOTE');
... ...
Remote SQL Information (identified by operation id):
----------------------------------------------------
3 - SELECT "OWNER","TABLE_NAME" FROM "T_TABLES" "A2" (accessing '!' )
Note
-----
- fully remote statement
语法：REMOTE_MAPPED([])
描述：在分布式查询中，指示优化器选择那个数据库的进行远程映射。作用和DRIVING_SITE类似。未指定参数时，采用本地数据库。
HELLODBA.COM>exec sql_explain('select /*+ remote_mapped(ORA11R2) */count(*) from T_USERS@ORA11R2 u, t_tables t where t.owner=u.username', 'TYPICAL');
Remote SQL Information (identified by operation id):
----------------------------------------------------
3 - SELECT "OWNER" FROM "T_TABLES" "A1" (accessing '!' )
Note
-----
- fully remote statement
语法：OPAQUE_TRANSFORM
描述：在分布式查询中，使用INSERT ... SELECT ... FROM语句从远程数据库查询数据插入本地数据库时，在远程数据库上执行的递归查询语句上会加上该提示，使得分布式数据库之间的兼容类型数据被透明传输。
(本地数据库，10.2.0.4)
HELLODBA.COM>create table t_objects_dummy2 as select * from t_objects@ora11r2 where 1=2;
Table created.
HELLODBA.COM>exec sql_explain('insert into t_objects_dummy2 select * from t_objects@ora11r2','TYPICAL');
… …
Remote SQL Information (identified by operation id):
----------------------------------------------------
1 - SELECT /*+ OPAQUE_TRANSFORM */ "OWNER","OBJECT_NAME","SUBOBJECT_NAME","OBJECT_I
D","DATA_OBJECT_ID","OBJECT_TYPE","CREATED","LAST_DDL_TIME","TIMESTAMP","STATUS","TEMP
ORARY","GENERATED","SECONDARY","NAMESPACE","EDITION_NAME","LIO" FROM "T_OBJECTS"
"T_OBJECTS" (accessing 'ORA11R2' )
HELLODBA.COM>insert into t_objects_dummy2 select * from t_objects@ora11r2;
72116 rows created.
(远程数据库，11.2.0.1)
HELLODBA.COM>select sql_text from v$sqlarea where sql_text like '%OPAQUE_TRANSFORM%' and sql_text not like '%v$sqlarea%';
SQL_TEXT
-------------------------------------------------------------------------------------------------------
SELECT /*+ OPAQUE_TRANSFORM */ "OWNER","OBJECT_NAME","SUBOBJECT_NAME","OBJECT_ID","DATA_OBJECT_ID","OBJECT_TYPE","CREATE
D","LAST_DDL_TIME","TIMESTAMP","STATUS","TEMPORARY","GENERATED","SECONDARY","NAMESPACE","EDITION_NAME","LIO" FROM "T_OBJECTS" "T_OBJECTS"
语法：STATEMENT_QUEUING
描述：在自动并行度模式(11gR2特性)下，使语句在并行资源不足时，进入等待队列，等到能获取到资源时继续运行；
自动并行度模式下，当存在多个并行查询同时在运行时，如果并行资源不足那么加了该提示的语句就会进入队列等待，此时，通过监控视图V$SQL_MONITOR可以看到其状态为QUEUE。
语法：NO_STATEMENT_QUEUING
描述：在自动并行度模式(11gR2特性)下，即使在并行资源不足时，也继续运行语句，这就可能会导致其它资源(如CPU)的争用与等待；
当参数“_parallel_statement_queuing”被设置为TRUE(默认为FALSE)，只有加上该提示的语句在并行资源紧张时不会进入队列。
语法：GBY_PUSHDOWN([])
描述：指示优化器在对并行查询进行代价估算时，考虑将GROUP BY操作推入并行服务进程的情况；
HELLODBA.COM>exec sql_explain('SELECT /*+ FULL(T) parallel(T DEFAULT) GBY_PUSHDOWN */ owner, table_name, COUNT (status) cnt FROM t_tables t GROUP BY owner, table_name', 'BASIC OUTLINE');
---------------------------------------------
| Id| Operation| Name|
---------------------------------------------
|0 | SELECT STATEMENT||
|1 |PX COORDINATOR ||
|2 |PX SEND QC (RANDOM)| :TQ10001 |
|3 |HASH GROUP BY||
|4 |PX RECEIVE||
|5 |PX SEND HASH| :TQ10000 |
|6 |HASH GROUP BY||
|7 |PX BLOCK ITERATOR ||
|8 |TABLE ACCESS FULL| T_TABLES |
---------------------------------------------
语法：NO_GBY_PUSHDOWN([])
描述：禁止优化器在对并行查询进行代价估算时，将GROUP BY操作推入并行服务进程；
HELLODBA.COM>exec sql_explain('SELECT /*+ FULL(T) parallel(T DEFAULT) NO_GBY_PUSHDOWN */ owner, table_name, COUNT (status) cnt FROM t_tables t GROUP BY owner, table_name', 'BASIC OUTLINE');
--------------------------------------------
| Id| Operation| Name|
--------------------------------------------
|0 | SELECT STATEMENT||
|1 |PX COORDINATOR||
|2 |PX SEND QC (RANDOM)| :TQ10001 |
|3 |HASH GROUP BY||
|4 |PX RECEIVE||
|5 |PX SEND HASH| :TQ10000 |
|6 |PX BLOCK ITERATOR ||
|7 |TABLE ACCESS FULL| T_TABLES |
--------------------------------------------
语法：HWM_BROKERED
描述：提示语句执行器在执行并行插入数据(或从其它表获取数据创建新表)时，使用高水位线查封器拆分高水位线，使得多个并行服务进程能共用一个扩展段。
示例(9i)：
HELLODBA.COM>alter session enable parallel dml;
Session altered.
HELLODBA.COM>explain plan for insert /*+ append */ into t_objects_dummy select /*+ full(o) parallel(o 2)*/* from t_objects o;
Explained.
HELLODBA.COM>select plan_table_output from table(dbms_xplan.display(null, null, 'ALL'));
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------
| Id| Operation|Name| Rows| Bytes | Cost|TQ|IN-OUT| PQ Distrib |
-------------------------------------------------------------------------------------------------
|0 | INSERT STATEMENT|| 32435 |2945K|22 ||||
|1 |LOAD AS SELECT||||||||
|2 |TABLE ACCESS FULL| T_OBJECTS| 32435 |2945K|22 | 63,00| P->S | QC (RAND)|
-------------------------------------------------------------------------------------------------
HELLODBA.COM>explain plan for insert /*+ append HWM_BROKERED */ into t_objects_dummy select /*+ full(o) parallel(o 2)*/* from t_objects o;
Explained.
HELLODBA.COM>select plan_table_output from table(dbms_xplan.display(null, null, 'ALL'));
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
-------------------------------------------------------------------------------------------------
| Id| Operation|Name| Rows| Bytes | Cost|TQ|IN-OUT| PQ Distrib |
-------------------------------------------------------------------------------------------------
|0 | INSERT STATEMENT|| 32435 |2945K|22 ||||
|1 |LOAD AS SELECT||||||||
|2 |TABLE ACCESS FULL| T_OBJECTS| 32435 |2945K|22 | 64,00| P->S | QC (RAND)|
|2 |TABLE ACCESS FULL| T_OBJECTS| 32435 |2945K|22 | 63,00| P->S | QC (RAND)|
|1 |LOAD AS SELECT||||||||
|2 |TABLE ACCESS FULL| T_OBJECTS| 32435 |2945K|22 | 64,00| P->S | QC (RAND)|
|2 |TABLE ACCESS FULL| T_OBJECTS| 32435 |2945K|22 | 63,00| P->S | QC (RAND)|
|0 | INSERT STATEMENT|| 32435 |2945K|22 ||||
|1 |LOAD AS SELECT||||||||
|2 |TABLE ACCESS FULL| T_OBJECTS| 32435 |2945K|22 | 64,00| P->S | QC (RAND)|
|2 |TABLE ACCESS FULL| T_OBJECTS| 32435 |2945K|22 | 63,00| P->S | QC (RAND)|
|1 |LOAD AS SELECT||||||||
|2 |TABLE ACCESS FULL| T_OBJECTS| 32435 |2945K|22 | 64,00| P->S | QC (RAND)|
|2 |TABLE ACCESS FULL| T_OBJECTS| 32435 |2945K|22 | 63,00| P->S | QC (RAND)|
-------------------------------------------------------------------------------------------------
语法：NO_QKN_BUFF
描述：禁止优化器使用动态分配的内存
HELLODBA.COM>exec sql_explain('select /*+parallel(t 8) parallel(o 8) leading(t o) pq_distribute(o hash hash) NO_QKN_BUFF*/* from t_tables t, t_objects o where t.owner=o.owner and t.table_name=o.object_name and o.status=:A','BASIC');
---------------------------------------------
| Id| Operation| Name|
---------------------------------------------
|0 | SELECT STATEMENT||
|1 |PX COORDINATOR||
|2 |PX SEND QC (RANDOM)| :TQ10002|
|3 |HASH JOIN||
|4 |PX RECEIVE||
|5 |PX SEND HASH| :TQ10000|
|6 |PX BLOCK ITERATOR ||
|7 |TABLE ACCESS FULL| T_TABLES|
|8 |PX RECEIVE||
|9 |PX SEND HASH| :TQ10001|
|10 |PX BLOCK ITERATOR ||
|11 |TABLE ACCESS FULL| T_OBJECTS |
---------------------------------------------
语法：PARALLEL_INDEX([查询块] [ ...] )或者PARALLEL_INDEX([查询块] () [() ...] )
描述：指示优化器选择并行方式访问本地分区索引。并行度可以为数字，也可以为DEFAULT，使用系统默认并行度。
HELLODBA.COM>exec sql_explain('select /*+ qb_name(M) parallel_index(o t_objects_list_IDX1 2) */* from t_objects_list o where object_name like :A', 'BASIC');
--------------------------------------------------------------------
| Id| Operation| Name|
--------------------------------------------------------------------
|0 | SELECT STATEMENT||
|1 |PX COORDINATOR||
|2 |PX SEND QC (RANDOM)| :TQ10000|
|3 |PX PARTITION LIST ALL||
|4 |TABLE ACCESS BY LOCAL INDEX ROWID| T_OBJECTS_LIST|
|5 |INDEX RANGE SCAN| T_OBJECTS_LIST_IDX1 |
--------------------------------------------------------------------
语法：NO_PARALLEL_INDEX([查询块] [ ...])或者NO_PARALLEL_INDEX([查询块] () [() ...])
描述：禁止优化器选择并行方式访问本地分区索引
HELLODBA.COM>alter index t_objects_list_IDX1 parallel(degree 2);
Index altered.
HELLODBA.COM>exec sql_explain('select /*+ qb_name(M) no_parallel_index(o t_objects_list_IDX1)*/* from t_objects_list o where object_name like :A', 'BASIC');
------------------------------------------------------------------
| Id| Operation| Name|
------------------------------------------------------------------
|0 | SELECT STATEMENT||
|1 |PARTITION LIST ALL||
|2 |TABLE ACCESS BY LOCAL INDEX ROWID| T_OBJECTS_LIST|
|3 |INDEX RANGE SCAN| T_OBJECTS_LIST_IDX1 |
------------------------------------------------------------------
语法：PQ_DISTRIBUTE([] )或者PQ_DISTRIBUTE([] )
描述：指定并行查询中并行收、发进程直接的分发方式；
HELLODBA.COM>exec sql_explain('select /*+parallel(o 2)*/* from t_objects o where exists (select /*+hash_sj PQ_DISTRIBUTE(t HASH HASH)*/1 from t_tables t where o.owner = t.owner and o.object_name = t.table_name)', 'BASIC');
-------------------------------------------------------
| Id| Operation| Name|
-------------------------------------------------------
|0 | SELECT STATEMENT||
|1 |PX COORDINATOR||
|2 |PX SEND QC (RANDOM)| :TQ10002|
|3 |HASH JOIN RIGHT SEMI BUFFERED||
|4 |BUFFER SORT||
|5 |PX RECEIVE||
|6 |PX SEND HASH| :TQ10000|
|7 |INDEX FULL SCAN| T_TABLES_PK |
|8 |PX RECEIVE||
|9 |PX SEND HASH| :TQ10001|
|10 |PX BLOCK ITERATOR||
|11 |TABLE ACCESS FULL| T_OBJECTS|
-------------------------------------------------------
语法：PARALLE([[] ] [])
描述：指示优化器对查询块或者对象使用并行查询。其中，当中指定整条语句为并行查询(未指定查询块和表)时，并行度可以为MANUAL,AUTO,DEFAULT或者指定数字；指定某个表时，并行度可以为DEFAULT或者指定数字，
HELLODBA.COM>exec sql_explain('SELECT /*+ parallel(u default) */* from t_users u', 'BASIC');
-----------------------------------------
| Id| Operation| Name|
-----------------------------------------
|0 | SELECT STATEMENT||
|1 |PX COORDINATOR||
|2 |PX SEND QC (RANDOM)| :TQ10000 |
|3 |PX BLOCK ITERATOR ||
|4 |TABLE ACCESS FULL| T_USERS|
-----------------------------------------
语法：NO_PARALLEL()
描述：禁止优化器并行查询表
HELLODBA.COM>alter table t_objects_dummy parallel(degree 2);
Table altered.
HELLODBA.COM>exec sql_explain('SELECT /*+ no_parallel */* from t_objects_dummy o', 'BASIC');
Plan hash value: 2093122083
---------------------------------------------
| Id| Operation| Name|
---------------------------------------------
|0 | SELECT STATEMENT||
|1 |TABLE ACCESS FULL| T_OBJECTS_DUMMY |
---------------------------------------------
语法：SHARED( [并行度])
描述：指示优化器共享指定表的并行度。如果指定并行度，则和PARALLEL提示作用相同。
HELLODBA.COM>alter table t_objects_dummy parallel(degree 2);
Table altered.
HELLODBA.COM>alter table t_objects parallel(degree 8);
Table altered.
HELLODBA.COM>exec sql_explain('select /*+ full(o) full(d) shared(d) */count(*) from t_objects_dummy d, t_objects o where o.owner=d.owner and o.object_name = d.object_name', 'BASIC COST');
------------------------------------------------------------------
| Id| Operation| Name| Cost (%CPU)|
------------------------------------------------------------------
|0 | SELECT STATEMENT||719(1)|
|1 |SORT AGGREGATE|||
|2 |PX COORDINATOR|||
|3 |PX SEND QC (RANDOM)| :TQ10002||
|4 |SORT AGGREGATE|||
| 5 |HASH JOIN||719(1)|
|6 |PX RECEIVE||231(0)|
|7 |PX SEND HASH| :TQ10000|231(0)|
|8 |PX BLOCK ITERATOR ||231(0)|
|9 |TABLE ACCESS FULL| T_OBJECTS|231(0)|
|10 |PX RECEIVE||486(0)|
|11 |PX SEND HASH| :TQ10001|486(0)|
|12 |PX BLOCK ITERATOR ||486(0)|
|13 |TABLE ACCESS FULL| T_OBJECTS_DUMMY |486(0)|
------------------------------------------------------------------
HELLODBA.COM>exec sql_explain('select /*+ full(o) full(d) shared(o) */count(*) from t_objects_dummy d, t_objects o where o.owner=d.owner and o.object_name = d.object_name', 'BASIC COST');
------------------------------------------------------------------
| Id| Operation| Name| Cost (%CPU)|
------------------------------------------------------------------
|0 | SELECT STATEMENT||1437(1)|
|1 |SORT AGGREGATE|||
|2 |PX COORDINATOR|||
|3 |PX SEND QC (RANDOM)| :TQ10002||
|4 |SORT AGGREGATE|||
|5 |HASH JOIN||1437(1)|
|6 |PX RECEIVE||463(1)|
|7 | PX SEND HASH| :TQ10000|463(1)|
|8 |PX BLOCK ITERATOR ||463(1)|
|9 |TABLE ACCESS FULL| T_OBJECTS|463(1)|
|10 |PX RECEIVE||973(1)|
|11 |PX SEND HASH| :TQ10001|973(1)|
|12 |PX BLOCK ITERATOR ||973(1)|
|13 |TABLE ACCESS FULL| T_OBJECTS_DUMMY |973(1)|
------------------------------------------------------------------
语法：NOPARALLEL([[] ])
描述：禁止优化器对查询块或者对象使用并行查询。和NO_PARALLEL作用基本相同。
参见NO_PARALLEL示例。
语法：PQ_MAP()
描述：未知。可能是用于并行查询的提示。
语法：PQ_NOMAP()
描述：未知。可能是用于并行查询的提示。
语法：PRESERVE_OID
描述：未知。可能是用于并行查询的提示。
语法：SYS_PARALLEL_TXN
描述：未知。可能是用于并行查询的递归调用语句上的。
语法：CUBE_GB
描述：未知。可能是用于GROUP BY CUBE并行查询的内部递归查询
该提示直接使用会导致10g(10.2.0.4)在解析提示时在后台发生ORA-00600错误，但不会终止语句运行。
ORA-600: internal error code, arguments: [prsHintQbLevel-1], [890], [], [], [], [], [], []
发生类似情况的提示还有：CUBE_GB/GBY_CONC_ROLLUP/PIV_GB/PIV_SSF/RESTORE_AS_INTERVALS/SAVE_AS_INTERVALS/SCN_ASCENDING/MODEL_DONTVERIFY_UNIQUENESS/TIV_GB/TIV_SSF
语法：GBY_CONC_ROLLUP
描述：未知。可能是用于GROUP BY ROLLUP并行查询的内部递归查询
语法：PIV_GB
描述：未知。出现在GROUP BY并行查询的内部递归查询语句上
示例(9i)：
HELLODBA.COM>select /*+qb_name(Q2) full(o2) parallel(o2 2)*/ owner, status, count(object_name) from t_objects o2 where owner like 'D%' group by owner, status;
Execution Plan
----------------------------------------------------------
0SELECT STATEMENT Optimizer=CHOOSE (Cost=11 Card=3 Bytes=42)
10SORT* (GROUP BY) (Cost=11 Card=3 Bytes=42):Q17865001
21SORT* (GROUP BY) (Cost=11 Card=3 Bytes=42):Q17865000
32TABLE ACCESS* (FULL) OF 'T_OBJECTS' (Cost=6 Card=3379 Bytes=47306):Q17865000
1 PARALLEL_TO_SERIALSELECT /*+ CIV_GB */ A1.C0,A1.C1,COUNT(SYS_O
P_CSR(A1.C2,0)) FROM :Q17865000 A1 GROUP BY
A1.C0,A1.C1
2 PARALLEL_TO_PARALLELSELECT /*+ PIV_GB */ A1.C0 C0,A1.C1 C1,SYS_O
P_MSR(COUNT(*)) C2 FROM (SELECT /*+ NO_EXPAN
D ROWID(A2) */ A2."OWNER" C0,A2."STATUS" C1
FROM "T_OBJECTS" PX_GRANULE(0, BLOCK_RANGE,
DYNAMIC)A2 WHERE A2."OWNER" LIKE 'D%') A1
GROUP BY A1.C0,A1.C1
3 PARALLEL_COMBINED_WITH_PARENT
语法：TIV_GB
描述：未知。出现在并行查询的内部递归查询语句上
语法：TIV_SSF
描述：未知。出现在并行查询的内部递归查询语句上
语法：PIV_SSF
描述：未知。出现在并行查询的内部递归查询语句上
语法：RESTORE_AS_INTERVALS
描述：未知。出现在并行查询的内部递归查询语句上
语法：SAVE_AS_INTERVALS
描述：未知。出现在并行查询的内部递归查询语句上
语法：SCN_ASCENDING
描述：未知。出现在并行查询的内部递归查询语句上
oracle ora 47306,Oracle SQL提示含义与示例 --- 分布式查询和并行提示相关推荐
- Oracle ora 01261,Oracle心得：ORA-01261:ORA-01262:错误的解决
用ps -ef|grep ora_ 查看没有发现Oracle的进程. 进入数据库启动遇上ORA-01261:ORA-01262:错误. $ sqlplus "/as sysdba" ...
- oracle ora 3136,oracle中ORA-3136,ORA-609
本文章详细的介绍了关于oracle中ORA-3136,ORA-609方法,有需要的朋友可以参考一下本文章哦. 本文章详细的介绍了关于oracle中ORA-3136,ORA-609方法,有需要的朋友可以 ...
- oracle ora 16014,ORACLE ORA-16014+ORA-00312 数据库在线重做日志没有归档
[oracle@RedHat ~]$ sqlplus / as sysdba SQL*Plus: Release 10.2.0.1.0 - Production on Fri Aug 30 11:23 ...
- oracle ora 24324,Oracle:ORA-24324: 未初始化服务句柄 ORA-24323: 不允许此值 ORA-01090: 正在关闭 – 不允许连接 | 学步园...
转自:http://blog.csdn.net/snoopy93/article/details/6234441 SQL> shutdown immediate ORA-24324: 未初始化服 ...
- 查询链接服务器信息,如何通过 SQL Server 链接服务器和分布式查询使用 Excel
作为链接服务器查询 Excel 您可以使用企业管理器.系统存储过程或 SQL-DMO(分布式管理对象)将 Excel 数据源配置为 SQL Server 链接服务器.在所有的这些情况中,您总需要设置以 ...
- oracle stdevp函数,ORACLE 系统函数与SQLSERVER系统函数的区别是什么
ORACLE 系统函数与SQLSERVER系统函数的区别是什么 发布时间:2021-03-17 14:40:36 来源:亿速云 阅读:57 作者:Leah 栏目:数据库 这篇文章给大家介绍ORACLE ...
- oracle 同义词名称,ORACLE同义词总结
同义词概念 Oracle的同义词(synonyms)从字面上理解就是别名的意思,和视图的功能类似,就是一种映射关系.它可以节省大量的数据库空间,对不同用户的操作同一张表没有多少差别;它扩展了数据库的使 ...
- ipguard客户端如何卸载_关于卸载Oracle步骤和相关SQL的学习
数据库相关概念 请解释什么是DB, DBMS, DBAS, DBA, DBS? DB:用于存储和管理数据的仓库 DBMS:用于科学的组织和存储数据, 高效快捷的管理和维护数据 DBAS:面向最终用户的 ...
- Oracle 9i与MS SQL Server 2000之比较连载五.zz
http://kb.cnblogs.com/a/1153156 Oracle 9i与MS SQL Server 2000之比较连载五 2008-04-14 18:13 四.Oracle中新的数据库对象 ...
最新文章
- ubuntu18.04.4 pip3 换源
- 什么才是有效的工程教育的方法？
- Git与Ftp协同工作
- 两个二进制数异或的结果
- dynamic 仪表板_仪表板完成百万美元交易
- 入门Vue.js要学习哪些知识？
- windows2k服务和进程详解
- 赚小钱，靠学一门技术没问题，那想赚大钱该怎么做呢？
- 一次大意引发的“大”故障
- 一部分 数据 迁移_yugong异构数据库迁移与同步
- 人生成功的十大说话技巧
- react native环境搭建的一些坑
- thinkphp5微信公众号推送模板消息
- DBMS Implementation 笔记 05: SIMC CATC 以及 Join 操作的实现
- 弱人工智能时代，如何跨越技术与产品的鸿沟？
- 记一次cocos逆向
- 电动车，摩托车灯照明 汽车灯照明 手电筒 驱动芯片
- 项目需求管理专栏︱如何进行高效的项目需求管理
- Frida不能和Magisk Hide一起使用
- 生成带大写英文字母和数字的验证码（手机或邮箱）