ORACLE1.16-练习题和索引在海量数据的应用
--思考：
--1.求算每个同学每科是否超过班级的单科平均分
-- 回顾：行转列
-- 1.利用decode（不只是用于行转列）
-- 2.利用oracle 11g以后提供pivot
select
ooo.*,
1 struts_avg,
2 srping_avg,
3 hibernate_avg,
4 oracle_avg
from (
select * from (
(select user_name,ke,score from t_user_score)
)
pivot(
sum(score) -- 统计值
for ke in ('struts','spring','hibernate','oracle') -- 划分多少列
)
) ooo
------------
-- struts的平均分
select * from t_user_score where ke='struts'
select avg(score) from t_user_score where ke='struts'
-------------------
select id ,
(select 1 from dual) pp
from t_user_score
--------------------------
select
ooo.*,
(select avg(score) from t_user_score where ke='struts') struts_avg,
(select avg(score) from t_user_score where ke='spring') srping_avg,
(select avg(score) from t_user_score where ke='hibernate') hibernate_avg,
(select avg(score) from t_user_score where ke='oracle') oracle_avg
from (
select * from (
(select user_name,ke,score from t_user_score)
)
pivot(
sum(score) -- 统计值
for ke in ('struts' as struts,'spring' as spring,'hibernate' as hibernate,'oracle' as oracle) -- 划分多少列
)
) ooo
--------------
select
ooo.*,
(select avg(score)-struts from t_user_score where ke='struts') low_struts_avg,
(select avg(score)-spring from t_user_score where ke='spring') low_srping_avg,
(select avg(score)-hibernate from t_user_score where ke='hibernate') low_hibernate_avg,
(select avg(score)-oracle from t_user_score where ke='oracle') low_oracle_avg
from (
select * from (
(select user_name,ke,score from t_user_score)
)
pivot(
sum(score) -- 统计值
for ke in ('struts' as struts,'spring' as spring,'hibernate' as hibernate,'oracle' as oracle) -- 划分多少列
)
) ooo
-------------------------------
select
ooo.*,
sign((select avg(score)-struts from t_user_score where ke='struts')) low_struts_avg,
sign((select avg(score)-spring from t_user_score where ke='spring')) low_srping_avg,
sign((select avg(score)-hibernate from t_user_score where ke='hibernate')) low_hibernate_avg,
sign((select avg(score)-oracle from t_user_score where ke='oracle')) low_oracle_avg
from (
select * from (
(select user_name,ke,score from t_user_score)
)
pivot(
sum(score) -- 统计值
for ke in ('struts' as struts,'spring' as spring,'hibernate' as hibernate,'oracle' as oracle) -- 划分多少列
)
) ooo
----------------------
--2.揪出偏科最严重的同学出来。
select
ooo.*,
decode(sign((select avg(score)-struts from t_user_score where ke='struts')),1,'是的，低于平均分',-1,'高于平均分','跟平均分持平') low_struts_avg,
decode(sign((select avg(score)-spring from t_user_score where ke='spring')),1,'是的，低于平均分',-1,'高于平均分','跟平均分持平') low_srping_avg,
decode(sign((select avg(score)-hibernate from t_user_score where ke='hibernate')),1,'是的，低于平均分',-1,'高于平均分','跟平均分持平') low_hibernate_avg,
decode(sign((select avg(score)-oracle from t_user_score where ke='oracle')),1,'是的，低于平均分',-1,'高于平均分','跟平均分持平') low_oracle_avg
from (
select * from (
(select user_name,ke,score from t_user_score)
)
pivot(
sum(score) -- 统计值
for ke in ('struts' as struts,'spring' as spring,'hibernate' as hibernate,'oracle' as oracle) -- 划分多少列
)
) ooo
-- 速度（提升速度）
-- 程序员必备技能
-- 1.绝对不能有not in
-- 2.对常见的查询条件的字段，需要添加索引
-- 3.对于海量的数据，设置分区
-- 系统维护
-- 保证数据库服务器内存足够大，让SGA足够大，保证缓存足够，减少硬盘的读写
-- 尽可能把最小而且查询频率最高的数据，要规划好表空间，放在SSD
--原理：
--索引，减少遍历（由头到尾的查询一次）
-- 2张表
create table t_user5 (
id number primary key,
user_name varchar2(100),
sex number
);
select key_value,display from dicts where key_type='sex' and lang='en'
insert into t_user5 values(1,'小平',1);
insert into t_user5 values(2,'小盛',1);
insert into t_user5 values(3,'小龙',0);
insert into t_user5 values(4,'小君',0);
insert into t_user5 values(5,'小芳',0);
commit
select * from t_user5
select key_value,display from dicts where key_type='sex' and lang='en'
-- 违反主键这个唯一性约束
insert into t_user5(select id,user_name,sex from t_user5)
select myseq.nextval from dual
insert into t_user5(
select
myseq.nextval,
user_name||id,
sex
from t_user5
)
commit
select count(1) from t_user5
-- 0.013 (利用主键唯一索引)
select * from t_user5 where id=8000000
--------------------
-- 0.014
select t_user5.id,t_user5.user_name,dicts.display from t_user5
left join dicts on
t_user5.sex=dicts.key_value
and key_type='sex' and lang='en'
where t_user5.id=8000000
------------------------
--- 2.193秒（相差了156倍）
select count(1) from t_user5
left join dicts on
t_user5.sex=dicts.key_value
and key_type='sex' and lang='en'
-- 耗时2.104
select count(1) from t_user5
left join dicts on
t_user5.sex=dicts.key_value
and key_type='sex' and lang='en'
where t_user5.user_name = '小平1'
------------------------------
-- 2.085
select * from t_user5
left join dicts on
t_user5.sex=dicts.key_value
and key_type='sex' and lang='en'
where t_user5.user_name = '小平1'
--- 数据从1000万提升（失败了）
-- 以空间换时间（写好书后，新建目录）
create index user5_name_index
on t_user5(user_name)
alter tablespace oa_data
add datafile 'c:\data\oa_data_003.dbf'
size 100m
autoextend on next 5m maxsize 2048m;
-----------------
--user_name 有了索引后，速度提升会0.045秒
select * from t_user5
left join dicts on
t_user5.sex=dicts.key_value
and key_type='sex' and lang='en'
where t_user5.user_name = '小平1'
select * from dicts
select * from t_user5
-------------------
insert into dicts (
select
myseq.nextval,
display,
myseq.nextval key_value,
key_type,
lang from dicts
)
commit
select count(1) from dicts
-- 没有索引：耗时1秒
select * from dicts where key_type='sex' and lang='en' and key_value=1
---------添加索引
create index dicts_kkl on dicts(key_value,key_type,lang)
-- 有索引：0.02秒
select * from dicts where key_type='sex' and lang='en' and key_value=1
