Other noteworthy remarks:

when mixing the money datatype with other datatypes, precision issues can occur.

http://stackoverflow.com/questions/582797/should-you-choose-the-money-or-decimalx-y-datatypes-in-sql-server

However, using the money data type can lead to faster SSAS processing performance.

I'm quite sure that SQL Server is much more secure than MS Access.

If your data is really sensitive, I'd avoid Access.

How did you move the 3 million records?

Check out this white paper:

The Data Loading Performance Guide

If you're worried about transaction logs, you might want to switch to bulk-logged recovery mode...

The error message is quite clear:

If you're the only user, try setting the RetainSameConnection property on the connection manager to true.

Btw, if you would like to use the REPLACE function, it should be something like this:

REPLACE([Description],[Description],City + ' - ' + [Zip-Code])

Does this work?

UPDATE c

SET [Description] = City + ' - ' + [Zip-Code]

FROM Client c

JOIN cityTable t ON c.[Description] = t.[Zip-Code]

What about the Attunity Teradata connector?

https://www.microsoft.com/en-us/download/details.aspx?id=44582

(you need Enterprise Edition though)

Great question. Never even realized EOMONTH had an optional second parameter. Good to know.

Nice question, thanks.

Thanks for the question.

Nice question. Quite tricky with computed columns, since they have restrictions as well.

RonKyle (9/13/2016)

What's the benefit of using integers in that context versus just using dates? For example, is there performance gains? Isn't date datatypes less bytes than int for one good...

Another good book:

Expert Performance Indexing

As a side note: I found the following book (aside from the Kimball books) an immensily good reference:

Star Schema - The Complete Reference

I did a review a while back:

Star Schema...

Certainly agree on not updating large, fast changing transactional fact tables.

Regarding the OP's question: I think the grain of the fact table is one order (which can contain many items),...