实体框架COUNT性能差

Question

We are experiencing very poor performance using Entity Framework 5.0 with MySql Connector 6.6.6.0 for count based queries. Our data structure looks like:

Table: Post
===========
ID           INT PRIMARY KEY
MemberID     INT NOT NULL
SiteID       INT NOT NULL
Description  VARCHAR(255) NOT NULL
Image        VARCHAR(255) NOT NULL
CreatedDate  DATETIME NULL

And using entity framework with a linq query like the following:

var count = entities.Post.Where(p => 
    p.SiteID == 1 && p.CreatedDate != null).Count();

We get the following generated SQL:

SELECT
`Extent1`.`ID`, 
`Extent1`.`MemberID`, 
`Extent1`.`SiteID`, 
`Extent1`.`Description`, 
`Extent1`.`Image`, 
`Extent1`.`CreatedDate`
FROM `Post` AS `Extent1`
 WHERE (`Extent1`.`SiteID` = 1) AND (`Extent1`.`CreatedDate` IS NOT NULL)

This reads all records and counts them in memory... Hugely inefficient as it should look something like:

SELECT COUNT(ID) FROM `Post` WHERE `SiteID` = 1 AND `CreatedDate` IS NOT NULL;

Is there anyway to hint to entity that we dont want to read all records into memory and just perform an SQL COUNT?

Answer 1

Try

var count = entities.Post.Where(p => 
       p.SiteID == 1 && p.CreatedDate != null).Query().Count();

http://msdn.microsoft.com/en-us/data/jj574232.aspx

Has this at the bottom of the page:

Using Query to count related entities without loading them

Sometimes it is useful to know how many entities are related to another entity in the database without actually incurring the cost of loading all those entities. The Query method with the LINQ Count method can be used to do this. For example:

using (var context = new BloggingContext()) 
{ 
    var blog = context.Blogs.Find(1); 

    // Count how many posts the blog has  
    var postCount = context.Entry(blog) 
                          .Collection(b => b.Posts) 
                          .Query() 
                          .Count(); 
}

Answer 2

Testing with EF 6

db.Users.Count(u => u.LastName == "xyz")

and

db.Users.Where(u=>u.LastName=="xyz").Count()

produce identical sql queries.

SELECT 
[GroupBy1].[A1] AS [C1]
FROM ( SELECT 
    COUNT(1) AS [A1]
    FROM [dbo].[Users] AS [Extent1]
    WHERE N'xyz' = [Extent1].[LastName]
)  AS [GroupBy1]