Question

I have a “server” table which has a column named ‘SN’ in mysql, when do query to retrive servers with some sns from ‘sn1’ to ‘sn10000’, we can:

select * from server where sn in ('sn1','sn2','sn3',...'sn10000');

If there is only one sn in ‘sn1’-‘sn10000’ which not exists in database, then the query above will retrive 9999 rows of result.

The question is how can I easily get which one in ‘sn1’-‘sn10000’ is not exists in database except the additional work, such as handling the result with shell script etc.

I have an ugly sql like below can use:

select * from (select 'sn1' as sn 
        union select 'sn2' 
        union select 'sn3'
        ....
        union select 'sn10000') as SN 
where not exists (select id from server where server.sn=SN.sn);

Is Anyone has other better methods? Thanks.

Answer 1

Your query is perfectly fine for the intended use, but on MySQL the NOT IN and LEFT JOIN/IS NULL are more effecient that NOT EXISTS. Here are your alternatives:

NOT IN

SELECT * 
  FROM (          SELECT 'sn1' as sn 
        UNION ALL SELECT 'sn2' 
        UNION ALL SELECT 'sn3'
        ....
        UNION ALL SELECT 'sn10000') as SN 
 WHERE sn.sn NOT IN (SELECT s.id FROM SERVER s)

LEFT JOIN/IS NULL

   SELECT s.id
     FROM SERVER s
LEFT JOIN (          SELECT 'sn1' as sn 
           UNION ALL SELECT 'sn2' 
           UNION ALL SELECT 'sn3'
           ....
           UNION ALL SELECT 'sn10000') as SN ON SN.sn = s.id
    WHERE sn.sn IS NULL

You might notice I used UNION ALL, rather than UNION – UNION removes duplicates (which won’t happen in your example), making it slower so UNION ALL is a better choice.