我正在使用带有LINQ实体的MySQL Connector / Net 6.5.4,并且我经常会遇到糟糕的查询性能,因为实体框架会生成使用派生表的查询。
这是我多次遇到的简化示例。在C#中,我写了一个这样的查询:
var culverCustomers = from cs in db.CustomerSummaries where cs.Street == "Culver" select cs;
// later...
var sortedCustomers = culverCustomers.OrderBy(cs => cs.Name).ToList();
而不是像这样生成简单的查询:
SELECT cust.id FROM customer_summary cust WHERE cust.street = "Culver" ORDER BY cust.name
实体框架使用派生表生成一个查询,如下所示:
SELECT Project1.id FROM (
SELECT cust.id, cust.name, cust.street FROM customer_summary cust
WHERE Project1.street = "Culver"
) AS Project1 -- here is where the EF generates a pointless derived table
ORDER BY Project1.name
如果我解释两个查询,我会在第一个查询中得到这个:
id, select_type, table, type, possible_keys, rows
1, PRIMARY, addr, ALL, PRIMARY, 9
1, PRIMARY, cust, ref, PRIMARY, 4
......对于实体框架查询而言,这样的事情很糟糕
id, select_type, table, type, possible_keys, rows
1, PRIMARY, <derived2>, ALL, 9639
2, DERIVED, addr, ALL, PRIMARY, 9
2, DERIVED, cust, ref, PRIMARY, 4
请注意第一行,其中MySQL解释说它正在扫描 9000+条记录。由于派生表,MySQL正在创建临时表并加载每一行。 (或者我根据这样的文章推断:Derived Tables and Views Performance)
如何阻止Entity Framework使用派生表,或者如何说服MySQL对此类查询进行明显的优化?
为了完成,以下是该linq查询的源视图:
create view customer_summary as
select cust.id, cust.name, addr.street
customers cust
join addresses addr
on addr.customer_id = cust.id
答案 0 :(得分:1)
我认为您的查询语句缺失了&#39; select&#39;。您尚未确定所需的记录。 您的查询:
var culverCustomers = from cs in db.CustomerSummaries
where cs.Street == "Culver";
//没有选择
你从桌子上选择什么?试试这个示例:
var culverCustomers = from cs in db.CustomerSummaries
where cs.Street == "Culver"
select cs;