can't force index

Hi,
I have a table called "users"
basically, it has a column called "user_id",
and there is a clustered index called "user_id_index",
so everytimes I try to execute
select * from users
I look at the execution plan, it's always used user_id_index, which is
fine.
however, I added another index called "user_name_index"
then I execute:
select * from users (index=user_name_index)
now I look at the execution plan, it is still using "user_id_index" ,
obviously,
it didn't force the index. what happen? why sql server ignore my index
hint?
how do you solve this problem?Recall that the clustered index holds the data as well as the key
columns. Unless the index "user_name_index" contains all the columns of
the table the server still has to read the clustered index to retrieve
the data. You should see a bookmark lookup on the cluster key.
Why do you see this as a problem? Why are you attempting to force an
index hint? Why are you using SELECT *, which potentially hinders index
optimization and shouldn't be used at all in production code.
David Portas
SQL Server MVP
--|||Hi
An not supplying a WHERE clause results in SQL server doing a table scan so
indexes may not be used (why use an index when you are returning all the
data?).
Regards
--
Mike Epprecht, Microsoft SQL Server MVP
Zurich, Switzerland
IM: mike@.epprecht.net
MVP Program: http://www.microsoft.com/mvp
Blog: http://www.msmvps.com/epprecht/
"David Portas" <REMOVE_BEFORE_REPLYING_dportas@.acm.org> wrote in message
news:1108762629.179991.114690@.o13g2000cwo.googlegroups.com...
> Recall that the clustered index holds the data as well as the key
> columns. Unless the index "user_name_index" contains all the columns of
> the table the server still has to read the clustered index to retrieve
> the data. You should see a bookmark lookup on the cluster key.
> Why do you see this as a problem? Why are you attempting to force an
> index hint? Why are you using SELECT *, which potentially hinders index
> optimization and shouldn't be used at all in production code.
> --
> David Portas
> SQL Server MVP
> --
>|||user_name_index is used on "user_name" column,
but even when I execute
select user_name from users (index=user_name_index)
where user_name='Joe'
I still see the execution plan is using clustered index "user_id_index".
don't you think it's weird?
"David Portas" <REMOVE_BEFORE_REPLYING_dportas@.acm.org> wrote in message
news:1108762629.179991.114690@.o13g2000cwo.googlegroups.com...
> Recall that the clustered index holds the data as well as the key
> columns. Unless the index "user_name_index" contains all the columns of
> the table the server still has to read the clustered index to retrieve
> the data. You should see a bookmark lookup on the cluster key.
> Why do you see this as a problem? Why are you attempting to force an
> index hint? Why are you using SELECT *, which potentially hinders index
> optimization and shouldn't be used at all in production code.
> --
> David Portas
> SQL Server MVP
> --
>|||> don't you think it's weird?
No. Did you read the replies from Mike and myself?
If you explain what you want to achieve maybe we can help you better.
If you just want to understand indexes and hints then I recommend Kalen
Delaney's book "Inside SQL Server". Hints are an advanced feature and
should be used only when essential and with an understanding of their
effects on the query plan.
David Portas
SQL Server MVP
--|||> select user_name from users (index=user_name_index)
> where user_name='Joe'
> I still see the execution plan is using clustered index
"user_id_index".
Maybe you could post some runnable code to reproduce that behaviour
(CREATE..., INSERT..., SELECT...). I don't see that myself. I get
user_name_index used with or without the hint. Also tell us your
edition, version and SP level.
David Portas
SQL Server MVP
--|||1) How big is the table? How many pages (blocks of 8 Kilobyte) does the
table use? You can check this with "sp_spaceused".
2) What version of SQL-Server are you using
3) Please post (simplified) DDL and some sample rows, and preferably a
script to reproduce the behavior.
Gert-Jan
Britney wrote:
> user_name_index is used on "user_name" column,
> but even when I execute
> select user_name from users (index=user_name_index)
> where user_name='Joe'
> I still see the execution plan is using clustered index "user_id_index".
> don't you think it's weird?|||Oh my god, sorry guys.
I was wrong about it, "users" is not a table, but it's a view.
I just found out.
In case you ask me why i'm doing this stupid view:
the reason we create a view for this is because I want to do snapshot
isolation for read and write.
if there are data coming in to [2users] table, then I alter view to use
[1users] table. So users table have 2 tables:
Read and write. This way I don't worry about locking.
CREATE VIEW users
AS
select * from [2users]
--
Now We know what is happening...
IF I select from actual table ,
select user_name from [2users] (index=user_name_index)
where user_name='Joe'
I see that it 's using forced index.
I guess view doesn't work correctly for some reason.
---
sp_spaceused [2users]
result:
name rows reserved data index_size unused
[2users] 41892 13952 KB 6616 KB 7144 KB 192 KB
---
select @.@.version
result:
Microsoft SQL Server 2000 - 8.00.780 (Intel X86) Mar 3 2003 10:28:28
Copyright (c) 1988-2003 Microsoft Corporation Enterprise Edition on Windows
NT 5.0 (Build 2195: Service Pack 4)
----

> 1) How big is the table? How many pages (blocks of 8 Kilobyte) does the
> table use? You can check this with "sp_spaceused".
>
"Gert-Jan Strik" <sorry@.toomuchspamalready.nl> wrote in message
news:42166EA6.EA6900C3@.toomuchspamalready.nl...
> 1) How big is the table? How many pages (blocks of 8 Kilobyte) does the
> table use? You can check this with "sp_spaceused".
> 2) What version of SQL-Server are you using
> 3) Please post (simplified) DDL and some sample rows, and preferably a
> script to reproduce the behavior.
> Gert-Jan
>
> Britney wrote: