我对hibernate和mybatis框架的比较

系统在选择操作数据库的框架上面，到底是选择hibernate,还是mybatis。 首先说下两者的原理，如果你要关联几张表做查询,查出20条记录: 1.如果是mybatis SELECT * FROM (SELECT INNER_TABLE.*, ROWNUM OUTER_TABLE_ROWNUM FROM (select SP_WORK_PLAN.name, spr

系统在选择操作数据库的框架上面，到底是选择hibernate,还是mybatis。

首先说下两者的原理，如果你要关联几张表做查询,查出20条记录:

1.如果是mybatis
SELECT *
FROM (SELECT INNER_TABLE.*, ROWNUM OUTER_TABLE_ROWNUM
FROM (select SP_WORK_PLAN.name, sprocorgan1_.code --只是查询
from SP_WORK_PLAN workplanvo0_,
v_sp_organization sprocorgan1_,
V_SP_USER sprocuserv2_,
v_sp_organization sprocorgan3_,
V_SP_USER sprocuserv4_,
V_SP_USER sprocuserv5_,
v_sp_organization sprocorgan6_
from workplanvo0_.APPLY_DEPARTMENT_OID =
sprocorgan1_.ORG_ID and
workplanvo0_.CONFIRMATION_UID = sprocuserv2_.USER_ID and
sprocuserv2_.ORG_ID = sprocorgan3_.ID and
workplanvo0_.CREATE_UID = sprocuserv4_.USER_ID and
workplanvo0_.WORK_MASTER_UID = sprocuserv5_.USER_ID and
workplanvo0_.WORK_TEAM_ID = sprocorgan6_.ORG_ID) INNER_TABLE
WHERE ROWNUM <= 20) OUTER_TABLE
WHERE OUTER_TABLE_ROWNUM > 0;

2.如果是hibernate,像下面的SQL要查20次。

select
workplanvo0_.ID as ID9_6_,
workplanvo0_.UPDATE_TIME as UPDATE2_9_6_,
workplanvo0_.OPTIMISTIC_LOCK_VERSION as OPTIMISTIC3_9_6_,
workplanvo0_.CODE as CODE9_6_,
workplanvo0_.DATA_FROM as DATA5_9_6_,
workplanvo0_.DATA_STATE as DATA6_9_6_,
workplanvo0_.END_LIFECYCLE as END7_9_6_,
workplanvo0_.NAME as NAME9_6_,
workplanvo0_.START_LIFECYCLE as START9_9_6_,
workplanvo0_.FLOW_STATE as FLOW10_9_6_,
workplanvo0_.PROCESS_INS_ID as PROCESS11_9_6_,
workplanvo0_.ACTUAL_END_TIME as ACTUAL12_9_6_,
workplanvo0_.ACTUAL_START_TIME as ACTUAL13_9_6_,
workplanvo0_.APPLY_DEPARTMENT_OID as APPLY40_9_6_,
workplanvo0_.ATTENTION_LEVEL as ATTENTION14_9_6_,
workplanvo0_.COMPLETE_CONDITION as COMPLETE15_9_6_,
workplanvo0_.CONFIRMATION_TIME as CONFIRM16_9_6_,
workplanvo0_.CONFIRMATION_UID as CONFIRM41_9_6_,
workplanvo0_.CREATE_UID as CREATE42_9_6_,
sprocorgan1_.ID as ID26_0_,
sprocorgan1_.UPDATE_TIME as UPDATE2_26_0_,
sprocorgan1_.OPTIMISTIC_LOCK_VERSION as OPTIMISTIC3_26_0_,
sprocorgan1_.CODE as CODE26_0_,
sprocorgan1_.DATA_FROM as DATA5_26_0_,
sprocorgan1_.DATA_STATE as DATA6_26_0_,
sprocorgan1_.END_LIFECYCLE as END7_26_0_,
sprocorgan1_.NAME as NAME26_0_,
sprocorgan1_.START_LIFECYCLE as START9_26_0_,
sprocorgan1_.AREA_ID as AREA10_26_0_,
sprocorgan1_.STATE as STATE26_0_,
sprocuserv2_.ID as ID27_1_,
sprocuserv2_.UPDATE_TIME as UPDATE2_27_1_,
sprocuserv2_.OPTIMISTIC_LOCK_VERSION as OPTIMISTIC3_27_1_,
sprocuserv2_.CODE as CODE27_1_,
sprocuserv2_.DATA_FROM as DATA5_27_1_,
sprocuserv2_.DATA_STATE as DATA6_27_1_,
sprocuserv2_.END_LIFECYCLE as END7_27_1_,
sprocuserv2_.NAME as NAME27_1_,
sprocuserv2_.START_LIFECYCLE as START9_27_1_,
sprocuserv2_.ACCOUNT as ACCOUNT27_1_,
sprocorgan3_.ID as ID26_2_,
sprocorgan3_.UPDATE_TIME as UPDATE2_26_2_,
sprocorgan3_.OPTIMISTIC_LOCK_VERSION as OPTIMISTIC3_26_2_,
sprocorgan3_.CODE as CODE26_2_,
sprocorgan3_.DATA_FROM as DATA5_26_2_,
sprocorgan3_.DATA_STATE as DATA6_26_2_,
sprocuserv4_.ID as ID27_3_,
sprocuserv4_.UPDATE_TIME as UPDATE2_27_3_,
sprocuserv4_.OPTIMISTIC_LOCK_VERSION as OPTIMISTIC3_27_3_,
sprocuserv4_.CODE as CODE27_3_,
sprocuserv4_.DATA_FROM as DATA5_27_3_,
sprocuserv4_.DATA_STATE as DATA6_27_3_,
sprocuserv4_.END_LIFECYCLE as END7_27_3_,
sprocuserv4_.NAME as NAME27_3_,
sprocuserv5_.ID as ID27_4_,
sprocuserv5_.UPDATE_TIME as UPDATE2_27_4_,
sprocuserv5_.OPTIMISTIC_LOCK_VERSION as OPTIMISTIC3_27_4_,
sprocuserv5_.CODE as CODE27_4_,
sprocuserv5_.DATA_FROM as DATA5_27_4_,
sprocuserv5_.DATA_STATE as DATA6_27_4_,
sprocuserv5_.END_LIFECYCLE as END7_27_4_,
sprocorgan6_.ID as ID26_5_,
sprocorgan6_.UPDATE_TIME as UPDATE2_26_5_,
sprocorgan6_.OPTIMISTIC_LOCK_VERSION as OPTIMISTIC3_26_5_,
sprocorgan6_.CODE as CODE26_5_,
sprocorgan6_.DATA_FROM as DATA5_26_5_,
sprocorgan6_.DATA_STATE as DATA6_26_5_,
sprocorgan6_.END_LIFECYCLE as END7_26_5_,
sprocorgan6_.NAME as NAME26_5_,
sprocorgan6_.START_LIFECYCLE as START9_26_5_,
sprocorgan6_.STATE as STATE26_5_
from
SP_WORK_PLAN workplanvo0_
left outer join
v_sp_organization sprocorgan1_ on workplanvo0_.APPLY_DEPARTMENT_OID=sprocorgan1_.ORG_ID
left outer join
V_SP_USER sprocuserv2_ on workplanvo0_.CONFIRMATION_UID=sprocuserv2_.USER_ID
left outer join
v_sp_organization sprocorgan3_ on sprocuserv2_.ORG_ID=sprocorgan3_.ID
left outer join
V_SP_USER sprocuserv4_ on workplanvo0_.CREATE_UID=sprocuserv4_.USER_ID
left outer join
V_SP_USER sprocuserv5_ on workplanvo0_.WORK_MASTER_UID=sprocuserv5_.USER_ID
left outer join
v_sp_organization sprocorgan6_ on workplanvo0_.WORK_TEAM_ID=sprocorgan6_.ORG_ID
where workplanvo0_.ID=? ;

1.设计阶段的影响
无法验证模型的合理性和预测性能。根据界面原型做数据库设计，优点是可以保证数据都能存到数据中，不足之处是无法保证模型的合理和性能。如果调整架构用ibatis，我们可以在设计完成后，写代码之前把复杂的查询写出来，制造一些数据进行性能预测。

2.开发阶段的影响
a.Hibernate无法使用层次查询、分析函数、正则表达式等。不错，hibernate有调用SQL的结果，如果通过接口调用SQL,不便于调试，用ibatis非常适合调试。
b.在开发阶段也无法预测性能,就是测试SQL。

c.hibernate的内部实现比较复杂，如果没有人能读懂里面的源代码，最好是只使用最简单的增、删、改，查（根据主键查）。反观mybatis，实现很简单，hold住。

3.运维阶段的影响
在运维阶段每天监控数据库，找到性能隐患，是业界的最佳实践。如果是hibernate的架构，即使我们找出了问题SQL，也无法对改动SQL进行调优，因为它是生成处理的，最多加一个索引。
最后：全部用hibernate对DB操作，把hibernate当做一个SQL生成的工具，其实就是把数据库当做一个黑盒，好像不需要对它有深入的了解，这样是被误导，看看我们现在有些组开发的报表开发工具，如果遇到大数据量，往往歇菜。