There's one big thing you are missing from your question - What are you trying to accomplish?  Changing the EXISTS clause to a join will cause you to have duplicate...

• This reply was modified 4 years, 7 months ago by pietlinden.

You have data. How about posting CREATE TABLE and INSERT scripts so we can see what you're working with? While you're at it, post the expected result... but for the...

Something like this? (I may be oversimplifying something...)

-- exclude when phone # refers to multiple people.

-- exclude when phone # refers to multiple people.
SELECT * FROM CustomerPolicies...

The easiest way is to use a Calendar table and explode the join

-- DROP TABLE #t
CREATE TABLE #t (CustNo int, Net money, StartDt datetime, EndDt datetime)
INSERT INTO...

• This reply was modified 4 years, 7 months ago by pietlinden.

use TOP() or LIMIT and do the delete in batches until @@ROWCOUNT = 0?

Post the CREATE TABLE scripts and someone can help you. =)

Great! Could you post the solution you came up with?

I'm no expert, but you have to unravel SQL  from the innermost queries to the outermost.

is there a shortcut? Not that I know of.  I'd love to be proven wrong,...

Your data is a bit wonky, but maybe something along these lines?

SELECT meshtype,

[datetime] as stamp,

tonsProduced,

theDate = CAST([datetime] AS DATE),

theTime = CAST([datetime] AS TIME)

-- fv = first_value([datetime]) OVER (PARTITION BY meshtype...

Normalization 101 time, I think. that's a LOT of repeating fields (any time I see a column name with a number at the end, I'm immediately suspect).

Is this supposed to...

• This reply was modified 4 years, 8 months ago by pietlinden.
• This reply was modified 4 years, 8 months ago by pietlinden.

This looks like it's supposed to be looking for the @current_proc_name value, not the literal string 'SOURCE', but then WTH do I know?

AND CHARINDEX('ALTER', @proc_source) < CHARINDEX('SOURCE',...

So what do you do then? Use XE to determine which queries are run and how often to determine whether the workload improves or not with the new indexing?

Create a unique index on all the columns and make that unique?

Multiply the result of your sum by 1.000 (adjust the number of zeroes as needed), and it will be converted to a decimal.

use datediff and calc full days first, then subtract, then repeat for hours, and then minutes?