一条HQL的变迁－－由584.14s->16.886s

事先申明，本人是菜鸟，口下留情。这个语句是用来计算负荷率的。计算的是每天的负荷率（峰电量＋谷电量＋平电量）／24／最高负荷的。先看看语句优化的过程，请高手解析这些语句的差异与优劣。
用词错误！！不叫优化，叫纠错（网友教的。。。）。。呵呵……菜鸟给个反面教材而已了。这些是我学习hibernate过程中弄出来的一个东西。。并非真正项目里的……。

(1)先看原始语句。

"select sum(model.peakQuantityElectricity+model.valleyQuantityElectricity+model.normalQuantityElectricity)/24/(dpss.wangGongFuHe+dpss.faDianFuHe)"
+" from DdRecord model,DatedPowerSupplyState dpss where model.recHost.hostType = 'SUPPLY_STATION' or model.recHost.hostType='POWER_PLANT' "
+"and model.date>=? and model.date<=? and dpss.date=model.date group by model.date"

执行耗时 584.12s，谢谢网友提示，这是个不对的hql语句。结果集大了很多，所以慢很多，应该将 model.recHost.hostType = 'SUPPLY_STATION' or model.recHost.hostType='POWER_PLANT' 部分括起来。
(2)先将中间or两个条件合并行，使用in语句（估计不好，好像见过说最好别用in的说法,不过我不知道如何改)。

"select sum(model.peakQuantityElectricity+model.valleyQuantityElectricity+model.normalQuantityElectricity)/24/(dpss.wangGongFuHe+dpss.faDianFuHe)"
+" from DdRecord model,DatedPowerSupplyState dpss where model.recHost.hostType  in ('SUPPLY_STATION' ,'POWER_PLANT') "
+"and model.date>=? and model.date<=? and dpss.date=model.date group by model.date"

居然有了很大提升，耗时17.5s
(3)想了想觉得DdRecord表中记录与DatedPowerSupplyState的记录数量的不同可能对语句有影响，试试调整下语句，调整如下:

"select sum(model.peakQuantityElectricity+model.valleyQuantityElectricity+model.normalQuantityElectricity)/24/(dpss.wangGongFuHe+dpss.faDianFuHe)"
+" from DdRecord model,DatedPowerSupplyState dpss where model.recHost.hostType  in ('SUPPLY_STATION' ,'POWER_PLANT')"
+" and dpss.date>=? and dpss.date<=? and model.date=dpss.date group by dpss.date"

执行后发现耗时17.7s，似乎没影响，或许还没有2的性能好。
(4)再想想觉得可以试试join 语句看看效果。调整后hql如下:

"select sum(model.peakQuantityElectricity+model.valleyQuantityElectricity+model.normalQuantityElectricity)/24/(dpss.wangGongFuHe+dpss.faDianFuHe)"
+" from DatedPowerSupplyState dpss,DdRecord model inner join model.recHost  rechost with rechost.hostType in ('POWER_PLANT','SUPPLY_STATION')"
+" where dpss.date>=? and dpss.date<=? and model.date=dpss.date group by dpss.date"

执行后，发现几乎没有影响，执行时间17.2s
(5)最后，综合了下，选择了以下语句:

"select sum(model.peakQuantityElectricity+model.valleyQuantityElectricity+model.normalQuantityElectricity)/24/(dpss.wangGongFuHe+dpss.faDianFuHe)"
+" from DatedPowerSupplyState dpss,DdRecord model inner join model.recHost  rechost with rechost.hostType in ('POWER_PLANT','SUPPLY_STATION')"
+" where model.date>=? and model.date<=? and dpss.date=model.date group by model.date"

耗时间为16.886s,结果还是不满意，最后解决办法多加了个表，每次有记录修改或新增时自动计算当日负荷率，存入这个表中。
综合上边过程提出一些问题：
（1） in语句和join语句使用对性能的影响。
（2）如果不取出这些记录，有无办法直接使用hql对查出来的结果作运算，比如select sqrt(select sum(...) from..) 之类。
（3）记录数量 对于编写hql 分组字段选取 或 条件语句编写等的影响。
恳请赐教。。另外给个语句，讨论下，如何优化:
// 站点名称

		"select rechost.name,"
				// 站点统计期内最高负荷
				+ "max(model.maximumLoad),"
				// 站点统计期内电量累计
				+ "sum(model.peakQuantityElectricity+model.valleyQuantityElectricity+model.normalQuantityElectricity),"
				// 站点统计期内总无功累计
				+ "sum(model.totalReactiveQuantityElectricity),"
				// 站点统计期内力率计算
				+ "sum((model.peakQuantityElectricity+model.valleyQuantityElectricity+model.normalQuantityElectricity)/sqrt((model.peakQuantityElectricity+model.valleyQuantityElectricity+model.normalQuantityElectricity)*(model.peakQuantityElectricity+model.valleyQuantityElectricity+model.normalQuantityElectricity)+model.totalReactiveQuantityElectricity*model.totalReactiveQuantityElectricity))/count(model.date)"
				+ " from RecordHost rechost,DdRecord model where model.recHost=rechost  and model.date>=? and model.date<=?"
				+ "group by rechost";