在我的系统中,我对具有大量集合的聚合进行集中计算。我需要在计算之前加载所有集合,因此我使用连接根集合的多标准。
我在下面列出的标准。在我的本地设置上运行大约需要500毫秒,这是很多用户的等待。有没有人对如何优化这个有任何想法?任何可以使这个查询更快的帮助都值得赞赏!
(我已经改变了一些隐私的集合名称:))
IMultiCriteria criteria = session.CreateMultiCriteria()
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("ACollection", JoinType.LeftOuterJoin)
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("BCollection", JoinType.LeftOuterJoin)
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("CCollection", JoinType.LeftOuterJoin)
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("ECollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("FCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("GCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("HCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("JCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("KCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("LCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("MCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("NCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("OCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("PCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("QCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("RCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("SCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("TCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("UCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("VCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("WCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("XCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("YCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("ZCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("AACollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("ABCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("ACCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("ADCollection", JoinType.LeftOuterJoin))
.Add(DetachedCriteria.For<Building>()
.Add(Restrictions.Eq("Id", BuildingId))
.CreateCriteria("AECollection", JoinType.LeftOuterJoin));
答案 0 :(得分:2)
过了一段时间,我终于意识到,对于这个确切的场景,使用像MongoDB或对象数据库这样的面向文档的数据库可能会更有意义。
这样我可以一次加载整个聚合并忘记连接。
因此,对于遇到上述情况的人,请考虑使用面向文档的方法。
我的初步测试表明了承诺:)
可以在此处找到MongoDB的一个介绍:http://mookid.dk/oncode/archives/1057
答案 1 :(得分:1)
我可以推荐的第一件事就是启动你的Sql Profiler并获得sql的精确副本。然后我会把它放在Sql Tuner中,这很可能会建议创建新的索引并在部分或全部表格上添加/更新统计数据。
此时我会进行性能测试,然后再尝试调整HQL的性能。