D'OH, got interrupted by work, forgot one thing.

Let's see if we can also reduce the number rows in the dates side of the JOIN. I'm assuming here that "Sunday"/"Saturday"...

Yeah, that join is not being processed well.

First, try just forcing a HASH join, i.e. writing "INNER HASH JOIN" instead of just INNER JOIN: that should be better here than...

swhetsell (11/19/2014)

---

ScottPletcher (11/19/2014)

---

Didn't see the test data when initially coding. I had to add RCN to the ORDER BY in case there are duplicate times ... although it seems...

Better would be an inline-table-valued-function.

If you intend to stick with a scalar function, at least get rid of the variable:

create function dbo.fu_salesTotal(@idSale int)

returns decimal(12,2)

as

begin;

return isnull((

...

Didn't see the test data when initially coding. I had to add RCN to the ORDER BY in case there are duplicate times ... although it seems odd/counterintuitive that...

Probably the big thing is to reduce all the gyrations around UnitLog. Please try the query below. If you tend to restrict UnitLog rows using fldDateTime range, particularly...

If you're padding a numeric value, I'd suggest using the built-in function:

STR()

I suspect that will perform better than writing the same thing out in your own code.

Btw, keep in mind that for multi-key-column indexes, SQL's missing index output does not consider the best order for the keys, which they freely admit. From Books Online, they...

The first thing to do with indexing is to review index usage stats and missing index stats together, at a minimum. If you're really serious, you also look at...

So is XXXXXXXXXXX by itself unique across all sites?

Or do you need the YYY and ZZZ values to get a unique key? Do you always specify YYY...

For this query:

select * from A where ReceiptTS > '2014-09-30 00:00:00.000'

SQL will use the actual date value to estimate the number of rows returned that will be returned.

For this query:

select...

measterbro (11/7/2014)

---

Since you want to criticize my table structure and imply that I am doing this for a school project and am therefore cheating, here is the actual table structure:

CREATE...

I think easiest would be to keep your own table of the original name and new name, then you can search for the original name in the history.

Do you have room on the target server for two copies of the db? If so, you have plenty of time, and can bring the second db fully current...

I used a CROSS APPLY to assign an alias name to the result of the string to datetime conversion, so the conversion doesn't have to be repeated anywhere.

I also changed...