如何在子选择中使用 Distinct 和 MAX

Question

I need to select distinct server_name, date with SUM of database_size, where I do a MAX of database_size in sub-select I can't manage to get distinct server_names.. my query is always returning duplicates (even though my table doesn't have duplicate values)

My table

id  server_name        database_size   received_at
1   abcd@ca.abc.com      271254016   2016-04-28
2   alphabet@ca.alp.com  61739776   2016-05-03
3   defg@ca.some.com     61438464   2016-05-03
4   1234@some.com        280043520  2016-05-03

The result I want: note* the date is distinct, the database_size is distinct and server_name is distinct-- none of these values should ever be the same.

abcd@ca.abc.com 271254016   2016-04-28
server2@ca.com  274209280   2016-04-29
server3@ca.com  268375040   2016-04-30
server4@ca.com  277126400   2016-05-01

I tried this query: This is not right, I want DISTINCT server_name (but putting DISTINCT doesn't help - it still returns same server_name)

select server_name, sum(data_size) as data_size, date from (select date(received_at) as date, sum(database_size) as data_size, 
server_name from schema.db_stats group by date(received_at), server_name) where  server_name like '%abc%' group by date, server_name order by date

But I get: I get server_name multiple times, I only want the server_name with the MAX amount of database_size

abcd@ca.abc.com 271254016   2016-04-28
abcd@ca.abc.com 274209280   2016-04-29
abcd@ca.abc.com 268375040   2016-04-30
abcd@ca.abc.com 277126400   2016-05-01
abcd@ca.abc.com 277126400   2016-05-02
alphabet@ca.alp.com 61739776    2016-05-03
defg@ca.some.com    61438464    2016-05-03
1234@some.com       280043520   2016-05-03

could someone please help ? Will really appreciate it.

Answer 1

I guess you are using Microsoft SQL Server, since you tagged your question sql . In the future, it would be helpful if you tag your questions with the brand of SQL database you use, because it will help people to answer with the best solution for that brand.

I have not tested this, but I believe something like the following will do what you want:

SELECT server_name, date, data_size
FROM (
    SELECT server_name, date,
      ROW_NUMBER() OVER (PARTITION BY server_name, date ORDER BY data_size DESC) AS RowNumber  
    FROM (
        SELECT server_name, DATE(received_at) AS date, SUM(database_size) as data_size,
        FROM schema.db_stats
        WHERE server_name LIKE '%abc%'
        GROUP BY server_name, DATE(received_at)
    )
)
WHERE RowNumber = 1;