You could run the job every minute.  But I'd say run the job every 10 minutes (or, I guess, 5 minutes).  Would it really be that big a deal to...

Are you using the standard sp_cycle_agent_errorlog proc or did you try something home grown? ...

In a single execution, the gain may be small, bring it up to 100K/Sec and that difference can...

Just a quick thought: Any IF statement will add an evaluation step to the execution Any AND will do the same Conditionals (IFs) introduce branching, even if there is...

I used a table for the location_id(s) to be combined, so that it would be easy to do multiple combinings.  For example, if you uncomment the (1, 30), you'll see...

In this specific case, I'd recommend you add a couple of triggers to the table to maintain an accurate flag for this in the parent row.

You'd need an INSERT trigger...

I agree, the first method is unnecessary.

As to whether I put all on one line: if they are straightforward, I would code them in one line with ANDs.  If one...

I really hope this isn't a dumb question. Is there a reason other then style or preference as to why you'd do this:

IF @Var1 =...

Unfortunately, we have a bit of a standard, which is that tables called "Thing" have a clustered index on ThingID, which will be an int, identity column. That's because...

The obvious difference is the number of rows read from the T_DeliveryTimetableItemResolvedCost table, which goes from 20k to 21 million (and a seek to a scan, probably unsurprisingly). The...

>>Use INSERT instead of MERGE in the second statement.

How would I then handle duplicates? I don't want to insert a duplicate record if there already is one that matches...

Since you only want monthly totals, I don't see any reason for the overhead to create a row for every single day; instead, you really only need a row for...

The SERIALIZABLE would cause higher levels of locks and cause them to be held longer.

Use INSERT instead of MERGE in the second statement.

Personally I'd use an UPDATE then INSERT on...

I guess the sequence must be called so many times that you're running out of values for an int.

Maybe use a bigint for the sequence?

I can't test these since you didn't provide directly usable test data:

--if you need only the acct_ids
SELECT acct_id, COUNT(*) OVER() AS total_accts
FROM my_table
GROUP BY acct_id
HAVING MIN(street +...