CELKO (10/20/2014)

---

..SEQUENCEs can "lose" numbers as well: if the transaction rolls back, the acquired sequence number(s) cannot be put back in the pool. If, by law, you can't have...

I prefer EXISTS over NOT IN especially, since NULL values will prevent NOT IN from working. You might see if that also corrects the optimizer issue.

select *

from table_A...

Yes, all modifications to recoverable resources for a single transaction must either all fail or all work, never some and not others. This is fundamental to how SQL Server...

With that many rows, you'll probably want to consider other, more efficient methods of comparing tables.

For example, change tracking would allow you to determine the rows in each table that...

SELECTc.cust_fullname AS Name,

c.cust_membership_id AS Account,

t.c_amount AS Amount,

'CH' + CAST(t.i_ticket_id AS varchar(12))...

In general, for a limited number of values (say 10 or...

My code to calculate max row length for a table -- and an index isn't that different -- shows 30 bytes, assuming 4 bytes for the clustering key.

The index you need for that DELETE would be keyed on:

( pk_Source, pk_cession )

If you only ever do this for pk_source = 2 (and not any other source numbers), you...

Try fso. It's reasonably fast if you don't go too crazy on the number of files. I left out the error checking, you can fill that in if...

Since the table has 1.5B rows, the ~860K rows accessed is only ~0.06%, so I can see why SQL chose to do (a lot of) SEEKs rather than a scan....

Probably the easiest way to do that is to create views which limit the data by date or whatever, then give them access to only the views. But that...

inevercheckthis2002 (10/16/2014)

---

Thank you, Scott.

So I take the same code and try to create a stored procedure by adding this:

IF OBJECT_ID('usp_DataLoadTime', 'P') IS NOT NULL

DROP PROC usp_DataLoadTime

GO

CREATE PROC usp_DataLoadTime

AS

BEGIN

... <code from...

The single biggest factor for performance is to get the best clustered index on every table.

Then adjust the non-clustered indexes as required.

Alexander Suprun (10/16/2014)

---

ScottPletcher (10/16/2014)

---

Alexander Suprun (10/16/2014)

---

Create a job which runs every 5 seconds and KILL the processes of specific user(s) where program_name is 'Microsoft SQL Server Management Studio - Query'...

Alexander Suprun (10/16/2014)

---

Create a job which runs every 5 seconds and KILL the processes of specific user(s) where program_name is 'Microsoft SQL Server Management Studio - Query' and that reached...