WITH (NOLOCK) = WITH (Occationally_Inconsistant_Data) 🙂

Nolock is something you use if you don't mind seeing dirty data, potantially missing rows and possibly reading rows twice. It's probably fine for long-running...

Duplicate post. Please do not reply to this.

http://www.sqlservercentral.com/Forums/Topic504304-291-1.aspx

SQL data and log files are not sparse. Snapshot data files are. Make sure you give your snapshot files a distinctive name (not .mdf) so you can easily distinguish them...

They extend the length of transactions, the increase the chance of deadlocks, and most people don't know how to write one. 😉

Do whatever you need to in the proc before...

No, we've just done a fair bit of performance tuning in the past and know where the issues usually lie and where the improvements...

As far as I can see, aa is just an alias for the openXML in the first query. It won't last longer than the execution of the first query. Hence,...

Clustered indexes should (imho) be as narrow as possible. From what you've said, I'd second Jeff and go for a cluster on A on just the ID. You can create...

Create indexes after the bcp/Bulk insert. That way they won't interfere with the load, but will be present when you're querying the table later.

As always, test carefully to ensure that...

A default just sets the value of the column if no...

I have one of them too, though it's a calendar mailing tube. A...

Glad to hear it.

Unfortunatly that's...

What we're saying is that you should only do hardware fixes if you know it's a hardware bottleneck, and you've already optimised the queries as much as possible and tuned...

GSquared (5/20/2008)

---

My thought was just put this logic in the proc that otherwise inserts into this table, and bypass the whole trigger and insert thing completely. But...

Can you post the execution plans please. You can right-click on the exec plan, save it as a .sqlplan file, zip and attach to the post. (assuming SQL 2005)

Somewhere I...