Well. Testing. So far it doesn't.

In some scenarios it does

But doing more testing now...

Thanks.

what is "destruction for equivalence"?

yes, I believe it can be, since at certain high contention times this SP is called 100 times a minute.

But once your statement is confirmed, what can possibly be done...

Yes, thanks. I got it.

Also, attached is the doc. that shows table structure, and a few typical calls to this SP (some of them yes, do happen within large(r) transactions), fired 100 times a...

Should I just replace  Update-->Insert-->Update (retry) with IF EXISTS   [update] ELSE [insert] ?

sorry; i intended to post in Oracle section but first posted here by mistake. How can I delete my post?

when the proc is called 100 times a minute or more, that's when deadlocks start. at something like 50 hits a minute or less, no deadlocks, only occasional blocking chains,...

Sorry I should have mentioned it. yes, it occasionally hits deadlocks.

Many times it is just blocking timeouts, a few times here and there there are deadlocks.

Thanks.

But how exactly do you suggest it can be re-written to do that?

Sorry. PDF file not allowed.

Attaching as WORD.

it has been pretty amazing for me to discover how few people know how little on issue(s) like this...

This is what I figured, too. Just wasn't sure..

So with that being said, can we assert that MS SQL Server had now an ADVANTAGE over Oracle in Columnstore technology usability/flexibility...

it is Windows 10 or Win Server 2016 OS.

Do you know where i can see an example of such triangular join?

thanks.