如何衡量休眠性能？

Question

如何衡量休眠中的性能？我想知道hibernate执行一个查询需要多少时间？

Answer 1

JProfiler 7.1 有一个 JPA/Hibernate 探针：

http://www.ej-technologies.com/products/jprofiler/whatsnew71.html

这是一个展示其工作原理的屏幕截图：

http://blog.ej-technologies.com/2012/01/profiling-jpahibernate.html

热点视图如下所示：

免责声明：我公司开发JProfiler

Answer 2

我会推荐以下一种或几种选择：

• 数据库探查器（例如 SqlServer Profiler）
• Hibernate 的内置统计信息（参见：http://www.javalobby.org/java/forums/t19807.html）
• Hibernate 分析器（参见：http://hibernateprofiler.com、http://nhprof.com/）

Answer 3

当我使用 Spring Hibernate 进行开发时 我通常使用以下属性
在我的 application.properties 文件中查看 查询执行时间和查询执行方式：

spring.jpa.properties.hibernate.show_sql=true
spring.jpa.properties.hibernate.use_sql_comments=true
spring.jpa.properties.hibernate.format_sql=true
spring.jpa.properties.hibernate.generate_statistics=true

示例日志输出如下所示：

Hibernate: 
select
    order0_.OrderID as OrderID1_4_0_,
    order0_.customerID as customer2_4_0_,
    order0_.employeeID as employee3_4_0_,
    order0_.freight as freight4_4_0_,
    order0_.orderDate as orderDat5_4_0_,
    order0_.requiredDate as required6_4_0_,
    order0_.shipAddress as shipAddr7_4_0_,
    order0_.shipCity as shipCity8_4_0_,
    order0_.shipCountry as shipCoun9_4_0_,
    order0_.shipName as shipNam10_4_0_,
    order0_.shipPostalCode as shipPos11_4_0_,
    order0_.shipRegion as shipReg12_4_0_,
    order0_.shipVia as shipVia13_4_0_,
    order0_.shippedDate as shipped14_4_0_,
    customer1_.customerID as customer1_2_1_,
    customer1_.address as address2_2_1_,
    customer1_.city as city3_2_1_,
    customer1_.companyName as companyN4_2_1_,
    customer1_.contactName as contactN5_2_1_,
    customer1_.contactTitle as contactT6_2_1_,
    customer1_.country as country7_2_1_,
    customer1_.fax as fax8_2_1_,
    customer1_.phone as phone9_2_1_,
    customer1_.postalCode as postalC10_2_1_,
    customer1_.region as region11_2_1_ 
from
    orders order0_ 
left outer join
    customers customer1_ 
        on order0_.customerID=customer1_.customerID 
where
    order0_.OrderID=?
Hibernate: 
select
    orderdetai0_.OrderID as OrderID6_3_0_,
    orderdetai0_.ID as ID1_3_0_,
    orderdetai0_.ID as ID1_3_1_,
    orderdetai0_.discount as discount2_3_1_,
    orderdetai0_.OrderID as OrderID6_3_1_,
    orderdetai0_.productID as productI3_3_1_,
    orderdetai0_.quantity as quantity4_3_1_,
    orderdetai0_.unitPrice as unitPric5_3_1_ 
from
    order_details orderdetai0_ 
where
    orderdetai0_.OrderID=?

Session Metrics {
281879 nanoseconds spent acquiring 1 JDBC connections;
0 nanoseconds spent releasing 0 JDBC connections;
276015 nanoseconds spent preparing 2 JDBC statements;
1702274 nanoseconds spent executing 2 JDBC statements;
0 nanoseconds spent executing 0 JDBC batches;
0 nanoseconds spent performing 0 L2C puts;
0 nanoseconds spent performing 0 L2C hits;
0 nanoseconds spent performing 0 L2C misses;
0 nanoseconds spent executing 0 flushes (flushing a total of 0 entities and 0 collections);
0 nanoseconds spent executing 0 partial-flushes (flushing a total of 0 entities and 0 collections)}

Answer 4

您是指字面意义上的查询还是查询 + 将数据转换为 Java 对象？

无论哪种方式，如果您想保持科学性，使用 JProbe 等分析工具可能是您的最佳选择。您不一定需要花钱，Eclipse 中也有一些免费工具。

在最简单的情况下，只要在有问题的代码周围添加一些打印语句就足够了，如果您要测量的是主导处理。

真正彻底地进行性能分析需要非常小心。通常，您需要多次重复您正在测试的东西，以确保您不会被初始开销（例如打开与 Db 的连接）误导。并且不要忘记数据库和性能可能很关键，hibernate 本身无法对调优的数据库做任何事情。