Brandie Tarvin (12/15/2010)

---

Paul White NZ (12/15/2010)

---

On the other hand, IN and EXISTS have slightly different NULL-handling semantics, so it is not possible to express the exact same logic using IN...

Thanks, Doodles 🙂

Yes and no. :pinch:

There would be no difference between the results from a query...

Stefan Krzywicki (12/14/2010)

---

Wow. I was having problems with an older query taking too long, so I made what I thought was a little switch to begin making time improvements. I...

Read the following blog entry by Conor Cunningham (Principal Software Architect, SQL Server Engine at Microsoft):

TOP 100 PERCENT...ORDER BY Considered Harmful

That explains everything you need to know about this issue.

For...

The last thing Craig needs right now is kids...

Craig,

For the deletes, use this:

DELETE psa

FROM #DelPortList AS dpl WITH (TABLOCKX)

JOIN PortSecAgg AS psa WITH (TABLOCKX)

ON psa.PortfolioID = dpl.PortfolioID

AND psa.AsOfDate = dpl.AsOfDate

OPTION...

I've written it both ways in different scenarios.

Wouldn't it be nice...

WayneS (12/14/2010)

---

Would there be any issues with building the staging table (dropping it first if it existed) in tempdb? This is a database that you can ensure will be in...

No worries. Have fun with it! I've enjoyed this thread, but I have some writing to do now. I'll stay subscribed to the thread just in case...

I think I'm going to tackle these one piece at a time. As strong as the partitioning structure sounds like it may be, I want to get baselines before I...

I'm curious as to why my existing clustered index on PortSecAgg, which is based AsOfDate/PortfolioID/hanging columns doesn't do the same thing, as a range removal. However, because each data file...

Having said I think it's best to keep things simple and skip the partition-switching with the delta loads, I see you asked about determining the partition number for the T1...

On the question of processing a single item at a time.

It should be clear that processing complete new-day files one at a time makes sense. By one at a...

Turning to your previous long post, Craig. First up, the sorting and column data types and nullability:

The idea of sorting the SSIS output on the clustered index keys of...