在ORACLE中的分析函数都是基于某几个字段划分计算窗口,然后在窗口内进行聚合,排名,等等计算。我想如果我们数据表的hash分区字段与分析函数中的partition by 字段一致的时候,应该可以大大加快分析函数的运行效率。因为每个分区上的数据可以单独进行运算。互不干涉,下面试验来验证我的想法.
第一步:创建一个分区表和普通表,表结构与DBA_OBJECTS一致:
create table t_partition_hash(object_name varchar2(128),subobject_name varchar2(30),object_id number,data_object_id number,object_type varchar2(19),created date,last_ddl_time date,timestamp varchar2(19),status varchar2(7),temporary varchar2(1),generated varchar2(1),secondary varchar2(1))partition by hash(object_type)(partition t_hash_p1 tablespace USERS,partition t_hash_p2 tablespace USERS,partition t_hash_p3 tablespace USERS,partition t_hash_p4 tablespace USERS,partition t_hash_p5 tablespace USERS,partition t_hash_p6 tablespace USERS,partition t_hash_p7 tablespace USERS,partition t_hash_p8 tablespace USERS); create table t_big_hash(object_name varchar2(128),subobject_name varchar2(30),object_id number,data_object_id number,object_type varchar2(19),created date,last_ddl_time date,timestamp varchar2(19),status varchar2(7),temporary varchar2(1),generated varchar2(1),secondary varchar2(1));
第二步:准备数据,从dba_object中把数据插入到两个表。总共插入数据1610880。
insert into t_partition_hash select * from dba_objects; insert into t_partition_hash select * from dba_objects;
第三步:本采用RANK函数对两个表进行查询。
begininsert into t_rankselect object_id, rank() over (partition by object_type order by object_id) r_object_id,rank() over (partition by object_type order by subobject_name) r_subobject_name ,rank() over (partition by object_type order by created) r_created,rank() over (partition by object_type order by last_ddl_time) r_last_ddl_time ,rank() over (partition by object_type order by status) r_object_typefrom t_partition_hash;end;
使用hash分区表总共执行5次的运行时间分别为:46.156s,33.39s,40.516s 34.875s 38.938s.
begininsert into t_rankselect object_id, rank() over (partition by object_type order by object_id) r_object_id,rank() over (partition by object_type order by subobject_name) r_subobject_name ,rank() over (partition by object_type order by created) r_created,rank() over (partition by object_type order by last_ddl_time) r_last_ddl_time ,rank() over (partition by object_type order by status) r_object_typefrom t_big_table;end; 使用非分区表执行5次的执行时间分别为:141.954s,89.656s,77.906s,98.5s,75.906s.
由此可见采用有效的HASH分区表可以有效提升分析函数在oracle中的执行效率。我相信随着数据量的增加,将会有更明显的效果,回头再测试一个项目中遇到的类似问题