Now that is a very interesting (serious) use case. I will have...

Ah, yes, I am responsible for performance, and that is just my frustration. These so-called developers are allowed to implement code which "works" on their test system with 16...

GilaMonster (12/29/2011)It's for when you don't mind long-lasting wide locks held by any operation until the end of the transaction, and when you have sufficient error handling and retry logic...

Sure! First come, first serve! 🙂

GSquared (12/29/2011)

patrickmcginnis59 (12/29/2011)

Brian.cs (12/29/2011)

RCSI really puts SQL Server running in a similar mode to what Oracle does by default, which is really what all robust modern RDBMS platforms should support....

Sadly, no, I do not have a test system where I can run my current environment under a load similar to production. (That would be awfully nice.)

I don't believe...

Create a stored procedure on your local server. Create the temp table you want inside that stored procedure. Call the remote stored procedure from within the local stored...

I recently experienced something similar. Two developers on our team could hardly get any work done in SSMS due to the horrible sluggishness in the application. No one...

This is the point when I need to step in and ask that we all remain positive and pleasant. 🙂

All of you are helping me by offering your technical opinions...

I don't think the estimates are that far off. A difference of 50% (or two to one) does not generally produce a bad execution plan. That's all you...

It is a puzzle. The first query delivers the performance I would expect with either query. The second query is loading a large portion of the table into...

It is hard to tell from the data you have provided. However, it appears that the second query (with the CTE) is actually a correlated sub-query.

(You join the CTE...

RCSI wouldn't permit dirty reads, but it would permit non-repeatable reads and phantom reads. Since the required overhead for the Temp DB is significant, I want to have some...

GSquared (12/28/2011)No, I agree with you. In a database that's loaded purely by a scheduled ETL process (or more than one), there's no reason to worry about dirty reads,...

David Portas makes a good point in his post. (I was following that thread until you guys started talking...