Physical on maxdop 5 or less behaves the same as the vm. 6 or up parallel.

But only NUMA 0 (8 cores) also ends up not parallel.

Anyway, found...

That's what I thought Grant - but then once the standard was on the new physical box, it still works.

Trimming the physical box to numa 0 only has made it...

Ah, no, I guess I wasn't clear.

It's the same data set on each server. Hence the confusion 🙂

I'm falling in the camp that says the query was wrong, with the table on the left being the wrong table.

If the query had been right it would have been...

"lessoning" of hardware failure.

Love it 🙂

Not sure that its really in the best of taste to do this on a release day... 🙂

Nicely written, thanks.

Add my voice to the concerns on this one...

as well as a +1 to Toreador for that comment.

Thanks for the replies, and the link.

It's looking like a SID lookup error, no AD related errors in the server logs though.

DBS all ok.

Elsewhere I found the hint to grant:

GRANT...

I would also firmly point you towards Sidney Dekkers Field Guide to understanding Human Error.

In my opinion one of the cross disciplinary must reads for any DBA.

Still a good article, but would be even better if you addressed some of the potential pitfalls of it.

Like runaway transaction log growth if you stop being able to clean...

All so much easier if you just spend an hour or 2 writing your own logshipping routines.

Is the change to (max) a biggie or just metadata? On say a 50GB table is it going to plough away at it, blocking the table for ages while...

Ok... so even stepping away from the question that is begging to be asked of why...

Would it not be better to add a bunch of 'or' additions, so that...

How about using Quotename...?

SELECT QUOTENAME('o''neil', '''')