Maybe I don't understand the question, but shouldn't you just modify the process that is running the backups so that it overwrites the file each night instead of generating a...

Yeah, that's an easy one:

UPDATE SysLog SET Hostname ='eub-dc-psesx-07'

from syslog

WHERE (DateTime >= 'Jan 2 2011 2:40:50:443AM')

AND (IP = '10.20.98.156') AND (Hostname = 'eub-dc-psesx-07') AND (EngineID = 1...

I know of this book. It's pretty good. It's on the MCM reading list... look down near the bottom of my post.

If you're working with SQL Server 2008, I'd go for using Change Data Capture. It's going to give you a lot more functionality, easily, than you can build yourself.

I am a huge fan of upping the cost threshold for parallelism. I'd set it higher, a lot higher, than the default for most OLTP systems I manage. 35 is...

Obviously, I want my cake and I want to eat it too (and I'll have as much of yours as I can)... or at least that's how most businesses would...

Indexes can make sort operations run faster. It does depend on what is being sorted, how many rows, what the rest of the query looks like etc. Try it. If...

r.pe (1/6/2011)

---

Sorry about that. I'll make this more concrete. Given a parent table A and a child table B, I can see multiple FKs defined for B referencing A on...

I don't completely understand. You're saying that if you ran a query against sys.foreign_keys and sys.foreign_key_columns that you would not see data returned if there are duplicate keys? That's not...

To my knowledge, you're in trouble when it comes to straight TSQL. There is no way to do this.

You could look at writing a CLR procedure to do it, but...

I'd change that statement to "sp_executesql can be parameterized" because queries run through sp_executesql...

EXEC and EXECUTE are the same, but the question was between EXEC...

the only way I know that you could do this is when your databases are running on a SAN. Then, you can use some of the transaction aware SAN utilities...

Yep, I'm recommending against it. I'm just using a lot of hedge words because you can make it work, but you probably shouldn't.

Just remember this, if you put all 11 columns into the clustered index, they are stored in every non-clustered index too. Also, if they're the primary key, they're in every...