Linq集团通过优化

Question

I have a mysql table, with lots of data, which looks like this:

fk_string CHAR(6) NOT NULL
timestamp BIGINT(20) NOT NULL

Note: there is a primary key defined that includes these two fields and there are other fields in the table, that will not be used in this example.

I would like for each fk_string, to get the total entry count, and the min and max timestamp. So I wrote the following Linq query:

var resQuery = db.Ticks.GroupBy(t => t.FkString)
    .Select(t => new {
        Key = t.Key
        , Total = t.LongCount()
        , First = t.Min(x => x.Timestamp)
        , Last = t.Max(x => x.Timestamp)
    });

The query generated by Linq is this one:

SELECT
    1 AS `C1`, 
    `GroupBy1`.`K1` AS `fk_string`, 
    `GroupBy1`.`A1` AS `C2`, 
    `GroupBy1`.`A2` AS `C3`, 
    `GroupBy1`.`A3` AS `C4`
FROM (SELECT
    `Project1`.`fk_string` AS `K1`, 
    COUNT(1) AS `A1`, 
    MIN(`Project1`.`C1`) AS `A2`, 
    MAX(`Project1`.`C1`) AS `A3`
    FROM (SELECT
        `Extent1`.`fk_string`, 
        `Extent1`.`timestamp` AS `C1`
        FROM `data` AS `Extent1`) AS `Project1`
    GROUP BY `Project1`.`fk_string`) AS `GroupBy1`

The problem is that because I'm reusing the timestamp column, Linq adds another subselect instead of just selecting directly from the table. The query I was expecting is this:

SELECT
    1 AS `C1`, 
    `GroupBy1`.`K1` AS `fk_string`, 
    `GroupBy1`.`A1` AS `C2`, 
    `GroupBy1`.`A2` AS `C3`, 
    `GroupBy1`.`A3` AS `C4`
FROM (SELECT
    `Project1`.`fk_string` AS `K1`, 
    COUNT(1) AS `A1`, 
    MIN(`Project1`.`C1`) AS `A2`, 
    MAX(`Project1`.`C1`) AS `A3`
    FROM `data` AS `Project1`
    GROUP BY `Project1`.`fk_string`) AS `GroupBy1`

and instead, the one generated by Linq takes 3x longer. Is there any way to disable or control this behaviour?

Answer 1

Perhaps moving the GroupBy into the Select ? And getting rid if Ticks?

Can you get LINQ to simply execute this:

SELECT fk_string AS FkString,
       COUNT(*) AS Total,
       MIN(timestamp) AS First,
       MAX(timestamp) AS Last,
       GROUP BY fk_string;

(without standing on your head to type the same stuff with extra keystrokes and function calls?)

Also, for performance:

INDEX(fk_string, timestamp)