select a.*

from table_a a

join

(

select

  loaded_unloaded,

  last_dt = max(scale_dt)

from table_a

  group by loaded_unloaded

)t

on a.loaded_unloaded = t.loaded_unloaded

and a.scale_dt = t.last_dt

Hi Erich,

1. How many pages does your index occury ?

2. Have you tried select * into some_table from your_fragmented_table with creating same inedexes on some_table?

select

sciid,

cityname,

rownum = when sciid = 4

 then rank() over (partition sciid order by cityname)

 else null

 end

from city

order by cityname

Yes, that's another reason why it's not 100% guarantee. However you can put

where text like '%c1%' or text like '%c2%' or text like '%c3%'

hope it will not return huge number...

Hi Yuri,

I had similar problem in a past, and here is how I solved it. However this method is not 100% guarantee that you will catch that SQL because it...

Hi Amit. Is it possible to do the same for grouping ?

select

* from sys.indexes

where

is_disabled = 1

alter

table emp add row_number int identity

go

delete

emp

from

...

It sounds like the problem is with design. To get only one record for each 100 and 101 I would recommend to normalize design of this table - to have...

Did you check your execution plan ?

What is the reason that optimizer does not use indexes ?

Is empty string is majority of records ?

Have statistics been updated recently...

The good book about 2005 which I like and highly recommend is Pro SQL Server 2005, by Thomas Rizzo, Apress.

Thanks Peter, I liked your solution. I already implemented it for my situation.

Yes Steve, it looks as it keeps entire string internally, but does not print it to output, even to a file.

So I will probably use methodology suggested by Chris, to print...

Hi Steve,

Yes, by the output I mean SSMS results window, but when I save it to a file it cuts this string too.

Here is quick script to test it:

set nocount...

You can also create a computed column where you will Checksum(your original column), then create a index on it.