as you have noticed, RI should have an index per FK-definition (unless the overlap is matching columnorder and sortdirection), otherwize indexes might not be used for RI-purposes.

Maybe use this to put them offline when sqlserver is started :

(you must be sysadmin to install this)

use master

create procedure sp_MyOwn_SetUserDbOffline

as

set nocount...

- are there any indexes on the foreign-key-columns to support the delete-join ? (with all child-tables)

- If yes, are the columns in the same columnorder and sortorder(asc/desc) as the parent...

keep in mind you are working with a heap table ! (indid =0 )

If you want to defrag a heap, you'll have to put a clustering index on it.

check BOL

create a dummy-db and transfert it to your SQL2k with the copy database wizard. It gives the possibility to migrate userid's.

dont just do it from the prod-db, because it performs...

I hope this gets you started :

SELECT  sysFile.groupid AS GroupID

, SUBSTRING(sysFile.groupname,1,30) AS FilegroupName

, SUBSTRING(sysObj.name,1,30) AS ObjectName

FROM sysobjects sysObj

 INNER JOIN sysindexes sysIdx

  ON sysObj.id = sysIdx.id

 INNER JOIN sysfilegroups sysFile...

What do you want to do ?

datepart(hh,**) will only return hours-part.

Datepart(mi,..) will only return minutes-part.

why not compare to convert(datetime, convert(char(10),getdate(),121) + ' 07:00:00'))

from Books Online :

DATEPART

Returns an integer...

Ad it to the comments on http://www.sqlservercentral.com/forums/shwmessage.aspx?forumid=32&messageid=11677 

- you could just have an alert raised when a file grows (e.g. error 5005 might be logged). Then extract the info from sql-log or windows-eventlog.

- you can gather the...

Correct.

no offence meant , just adding the symptoms.

I know that was the goal, but having a 48Kb table using +500Mb is what happened . After creating the CI sizes were back...

- Keep in mind that your clustering-index may be uniqueified by sqlserver itself if it is not a unique key _and_ it is added to all nci ! So keep...

stop it and start it again without the -m parameter

I'd suggest you make a stored proc to perform this functionality. Use temporary objects to store intermediate data and returen your selected subset from the temporary object.

store it as a separate smallint, just indicating timedifference to GMT. If doing so, it might be interesting to include a gmt-datetime column because otherwise you'll have to include "daylightsavings"...