Sql – Select statement performance

join;selectsql server

I'm having a performance issue with a select statement I'm executing.

Here it is:

SELECT     Material.*
FROM       Material 
INNER JOIN LineInfo ON Material.LineInfoCtr = LineInfo.ctr 
INNER JOIN Order_Header ON LineInfo.Order_HeaderCtr = Order_Header.ctr
WHERE   (Order_Header.jobNum = 'ttest') 
    AND (Order_Header.revision_number = 0) 
    AND (LineInfo.lineNum = 46)

The statement is taking 5-10 seconds to execute depending on server load.

Some table stats:

- Material has 2,030,xxx records.
- Lineinfo has 190,xxx records
- Order_Header has 2,5xx records.

My statement is returning a total of 18 rows containing about 20-25 fields of data. Returning a single field or all of them makes no difference. Is this performance typical? Is there something I could do to improve it?

I've tried using a sub select to retrieve the foreign key, the IN clause and I found one post where a fella said using a left outer join helped him. For me, they all yield the same 5 to 10 seconds of execution time.

This is MS SQL server 2005 accessed through MS SQL management studio. Times are the elapsed time in query analyzer.

Any ideas?

Best Answer

The first thing you should do is analyze the query plan, to see what indexes (if any) SQL Server is using.

You can probably benefit from some covering indexes in this query, since you only use columns in Lineinfo and Order_Header for the join and the query restriction (the WHERE clause).

Related Solutions

Sql – How to perform an IF…THEN in an SQL SELECT

The CASE statement is the closest to IF in SQL and is supported on all versions of SQL Server.

SELECT CAST(
             CASE
                  WHEN Obsolete = 'N' or InStock = 'Y'
                     THEN 1
                  ELSE 0
             END AS bit) as Saleable, *
FROM Product

You only need to use the CAST operator if you want the result as a Boolean value. If you are happy with an int, this works:

SELECT CASE
            WHEN Obsolete = 'N' or InStock = 'Y'
               THEN 1
               ELSE 0
       END as Saleable, *
FROM Product

CASE statements can be embedded in other CASE statements and even included in aggregates.

SQL Server Denali (SQL Server 2012) adds the IIF statement which is also available in access (pointed out by Martin Smith):

SELECT IIF(Obsolete = 'N' or InStock = 'Y', 1, 0) as Saleable, * FROM Product

Sql – How to do an UPDATE statement with JOIN in SQL Server

Syntax strictly depends on which SQL DBMS you're using. Here are some ways to do it in ANSI/ISO (aka should work on any SQL DBMS), MySQL, SQL Server, and Oracle. Be advised that my suggested ANSI/ISO method will typically be much slower than the other two methods, but if you're using a SQL DBMS other than MySQL, SQL Server, or Oracle, then it may be the only way to go (e.g. if your SQL DBMS doesn't support MERGE):

ANSI/ISO:

update ud 
     set assid = (
          select sale.assid 
          from sale 
          where sale.udid = ud.id
     )
 where exists (
      select * 
      from sale 
      where sale.udid = ud.id
 );

MySQL:

update ud u
inner join sale s on
    u.id = s.udid
set u.assid = s.assid

SQL Server:

update u
set u.assid = s.assid
from ud u
    inner join sale s on
        u.id = s.udid

PostgreSQL:

update ud
  set assid = s.assid
from sale s 
where ud.id = s.udid;

Note that the target table must not be repeated in the FROM clause for Postgres.

Oracle:

update
    (select
        u.assid as new_assid,
        s.assid as old_assid
    from ud u
        inner join sale s on
            u.id = s.udid) up
set up.new_assid = up.old_assid

SQLite:

update ud 
     set assid = (
          select sale.assid 
          from sale 
          where sale.udid = ud.id
     )
 where RowID in (
      select RowID 
      from ud 
      where sale.udid = ud.id
 );

Best Answer

Related Solutions

Sql – How to perform an IF…THEN in an SQL SELECT

Sql – How to do an UPDATE statement with JOIN in SQL Server

Related Topic