The difficulty you are having here is exactly why I said the meaning of 'set-based' is a bit woolly, in response to Sean saying that recursive CTEs are not set-based.

How...

Jeff Moden (5/25/2012)

---

It's certainly an unusual way of doing a cross join, but yes... I consider that to be set based. And, no, I don't consider it to be...

Yes, I appreciate that. My rhetorical aside was really just to myself; though it does get...

Not as often as it should? At least until the primary question has been answered.

Gullimeel (5/24/2012)

---

I am aware of that bug and min on ascending order index wont cause the issue but max will do.I could have very well used min but i...

Gullimeel (5/25/2012)

---

This is simple arithmatic that IO cost + cpu cost is the total operator cost.How optimizer is calculating IO cost or CPU cost that is explained very well in...

isuckatsql (5/25/2012)

---

I removed the reference to "AllRecentWages" and got 850k records in 10 seconds, so that was certainly a big part of the issue! Now i need to figure out...

RBarryYoung (5/25/2012)

---

I *really* wish that you would write a book with all of this stuff in it. Every time you post one of these I am just dumbstruck with...

Language issues,...

Gullimeel (5/25/2012)

---

In attached screenshot if you will see that the IO cost estimation is 185.2 and CPU estimation cost is around 4.8. Then total subtree cost should have been around...

Thanks. Perhaps the problem is ghost cleanup instead? The following link contains information about this, as...

Several parallelism operators in the parallel plan have a warning symbol to show you that exchange buffers were...

USE AdventureWorks2008R2;

GO

DECLARE @sql nvarchar(MAX) =

(

SELECT

N'DROP ' +

QUOTENAME(SCHEMA_NAME(v.[schema_id])) +

...

It seems likely the table is a heap (i.e. there is no clustered index). Empty pages are not automatically deallocated when deleting from a heap, unless a WITH (TABLOCK)...

I wish :Wow: