Click here to monitor SSC
SQLServerCentral is supported by Red Gate Software Ltd.
 
Log in  ::  Register  ::  Not logged in
 
 
 
        
Home       Members    Calendar    Who's On


Add to briefcase ««12

using multiple UNION in query Expand / Collapse
Author
Message
Posted Thursday, May 16, 2013 8:16 AM


SSCertifiable

SSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiable

Group: General Forum Members
Last Login: 2 days ago @ 2:06 PM
Points: 5,047, Visits: 11,797
Thanks Sean - that was the point that I was hoping to make.


Help us to help you. For better, quicker and more-focused answers to your questions, consider following the advice in this link.

When you ask a question (and please do ask a question: "My T-SQL does not work" just doesn't cut it), please provide enough information for us to understand its context.
Post #1453573
Posted Thursday, May 16, 2013 8:40 AM


SSCertifiable

SSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiable

Group: General Forum Members
Last Login: Yesterday @ 1:24 PM
Points: 5,318, Visits: 9,760
Sean Lange (5/16/2013)
John Mitchell-245523 (5/16/2013)
Phil Parkin (5/16/2013)
lnardozi 61862 (5/16/2013)
Just for completeness sake, you never use more than one UNION, which is the last one. All the rest should be UNION ALL.


Why not?

I suppose because UNION eliminates duplicates you only need to do it once - as long as you make sure that UNION is the last one evaluated. I don't know whether the query optimizer is smart enough to do that implicitly if you use more than one UNION - so it may or may not make a difference from a performance point of view.

John


But that is only good if you want to eliminate ALL duplicates. It is very possible that you have 3 tables and in the first 2 tables you want duplicates but if the same value shows up in the third table you want only distinct values. Other times maybe you want to include all duplicates.

The blanket statement that "you never use more than one UNION, which is the last one. All the rest should be UNION ALL." is just completely not true. There are time when that is correct but not always.

I'm not sure I understand. Those duplicates from the first two tables will be eliminated by the UNION operator. Try this:
SELECT * FROM (VALUES (1), (1), (2)) x(A)
UNION ALL
SELECT * FROM (VALUES (2), (3), (3)) y(B)
UNION
SELECT * FROM (VALUES (4)) z(C)

John
Post #1453585
Posted Thursday, May 16, 2013 8:57 AM
SSCrazy

SSCrazySSCrazySSCrazySSCrazySSCrazySSCrazySSCrazySSCrazy

Group: General Forum Members
Last Login: Tuesday, August 26, 2014 7:27 AM
Points: 2,836, Visits: 5,067
John Mitchell-245523 (5/16/2013)
Sean Lange (5/16/2013)
John Mitchell-245523 (5/16/2013)
Phil Parkin (5/16/2013)
lnardozi 61862 (5/16/2013)
Just for completeness sake, you never use more than one UNION, which is the last one. All the rest should be UNION ALL.


Why not?

I suppose because UNION eliminates duplicates you only need to do it once - as long as you make sure that UNION is the last one evaluated. I don't know whether the query optimizer is smart enough to do that implicitly if you use more than one UNION - so it may or may not make a difference from a performance point of view.

John


But that is only good if you want to eliminate ALL duplicates. It is very possible that you have 3 tables and in the first 2 tables you want duplicates but if the same value shows up in the third table you want only distinct values. Other times maybe you want to include all duplicates.

The blanket statement that "you never use more than one UNION, which is the last one. All the rest should be UNION ALL." is just completely not true. There are time when that is correct but not always.

I'm not sure I understand. Those duplicates from the first two tables will be eliminated by the UNION operator. Try this:
SELECT * FROM (VALUES (1), (1), (2)) x(A)
UNION ALL
SELECT * FROM (VALUES (2), (3), (3)) y(B)
UNION
SELECT * FROM (VALUES (4)) z(C)

John


try this:
SELECT * FROM (VALUES (1), (1), (2)) x(A)
UNION
SELECT * FROM (VALUES (2), (3), (3)) y(B)
UNION all
SELECT * FROM (VALUES (4),(3)) z(C)

However, until this sort of functionality is really required, I wouldn't mix both UNION and UNION ALL in one query.



_____________________________________________
"The only true wisdom is in knowing you know nothing"
"O skol'ko nam otkrytiy chudnyh prevnosit microsofta duh!"
(So many miracle inventions provided by MS to us...)

How to post your question to get the best and quick help
Post #1453602
Posted Thursday, May 16, 2013 9:05 AM


SSChampion

SSChampionSSChampionSSChampionSSChampionSSChampionSSChampionSSChampionSSChampionSSChampionSSChampion

Group: General Forum Members
Last Login: Yesterday @ 2:32 PM
Points: 13,302, Visits: 12,168
John Mitchell-245523 (5/16/2013)
Sean Lange (5/16/2013)
John Mitchell-245523 (5/16/2013)
Phil Parkin (5/16/2013)
lnardozi 61862 (5/16/2013)
Just for completeness sake, you never use more than one UNION, which is the last one. All the rest should be UNION ALL.


Why not?

I suppose because UNION eliminates duplicates you only need to do it once - as long as you make sure that UNION is the last one evaluated. I don't know whether the query optimizer is smart enough to do that implicitly if you use more than one UNION - so it may or may not make a difference from a performance point of view.

John


But that is only good if you want to eliminate ALL duplicates. It is very possible that you have 3 tables and in the first 2 tables you want duplicates but if the same value shows up in the third table you want only distinct values. Other times maybe you want to include all duplicates.

The blanket statement that "you never use more than one UNION, which is the last one. All the rest should be UNION ALL." is just completely not true. There are time when that is correct but not always.

I'm not sure I understand. Those duplicates from the first two tables will be eliminated by the UNION operator. Try this:
SELECT * FROM (VALUES (1), (1), (2)) x(A)
UNION ALL
SELECT * FROM (VALUES (2), (3), (3)) y(B)
UNION
SELECT * FROM (VALUES (4)) z(C)

John


The difference is that you don't have any values in z that are also in either of the first two tables.

Let's say you wanted distinct values from x and y but you wanted to include duplicates from z if they exist. In this scenario you have to change the order of the UNION and UNION ALL.

SELECT * FROM (VALUES (1), (1), (2)) x(A)
UNION
SELECT * FROM (VALUES (2), (3), (3)) y(B)
UNION ALL
SELECT * FROM (VALUES (4), (1)) z(C)

--EDIT--

Didn't see Eugene's post before I hit reply. I agree that you should not mix them unless it is really necessary. It does however prove that the blanket statement posted above is just not true. As with nearly everything in SQL "it depends".


_______________________________________________________________

Need help? Help us help you.

Read the article at http://www.sqlservercentral.com/articles/Best+Practices/61537/ for best practices on asking questions.

Need to split a string? Try Jeff Moden's splitter.

Cross Tabs and Pivots, Part 1 – Converting Rows to Columns
Cross Tabs and Pivots, Part 2 - Dynamic Cross Tabs
Understanding and Using APPLY (Part 1)
Understanding and Using APPLY (Part 2)
Post #1453608
Posted Thursday, May 16, 2013 9:08 AM


SSCertifiable

SSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiable

Group: General Forum Members
Last Login: Yesterday @ 1:24 PM
Points: 5,318, Visits: 9,760
But that's because you changed the order. The proviso was that the UNION operator is the last one to be evaluated. If you use parentheses to make it so, you get the same results as before:

SELECT * FROM (VALUES (1), (1), (2)) x(A)
UNION
(SELECT * FROM (VALUES (2), (3), (3)) y(B)
UNION all
SELECT * FROM (VALUES (4),(3)) z(C))

John

Edit - I think I understand what you're saying. I need to play around and see what happens when you have more than one UNION. I'll see if I get time to do that tomorrow.
Post #1453611
Posted Thursday, May 16, 2013 9:14 AM


SSChampion

SSChampionSSChampionSSChampionSSChampionSSChampionSSChampionSSChampionSSChampionSSChampionSSChampion

Group: General Forum Members
Last Login: Yesterday @ 2:32 PM
Points: 13,302, Visits: 12,168
John Mitchell-245523 (5/16/2013)
But that's because you changed the order. The proviso was that the UNION operator is the last one to be evaluated. If you use parentheses to make it so, you get the same results as before:

SELECT * FROM (VALUES (1), (1), (2)) x(A)
UNION
(SELECT * FROM (VALUES (2), (3), (3)) y(B)
UNION all
SELECT * FROM (VALUES (4),(3)) z(C))

John

Edit - I think I understand what you're saying. I need to play around and see what happens when you have more than one UNION. I'll see if I get time to do that tomorrow.


Right but they made a blanket statement that you should NEVER us more than one union and it should always be the last one. That is just simply false. It depends on the requirements of the result set.


_______________________________________________________________

Need help? Help us help you.

Read the article at http://www.sqlservercentral.com/articles/Best+Practices/61537/ for best practices on asking questions.

Need to split a string? Try Jeff Moden's splitter.

Cross Tabs and Pivots, Part 1 – Converting Rows to Columns
Cross Tabs and Pivots, Part 2 - Dynamic Cross Tabs
Understanding and Using APPLY (Part 1)
Understanding and Using APPLY (Part 2)
Post #1453616
Posted Friday, May 17, 2013 1:20 AM


SSCertifiable

SSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiable

Group: General Forum Members
Last Login: Yesterday @ 1:24 PM
Points: 5,318, Visits: 9,760
Yes, I agree that it was worded too strongly. Can you think of any cases where it isn't possible to rewrite a series of UNIONs so that it contains only one UNION, with the rest being converted to UNION ALLs?

John
Post #1453873
« Prev Topic | Next Topic »

Add to briefcase ««12

Permissions Expand / Collapse