Insert operations can absolutely block one another. If a lock, or a latch, is held on a page by one operation, then other operations will wait for access. Latches are...

All the DISTINCT operators... Either, your data structures are very poor, so you can't guarantee unique rows based on standard queries, or, someone thinks this is how all queries must...

and speaking of blocking.

whoever did the code does not really understand what "SET TRANSACTION ISOLATION LEVEL READ UNCOMMITTED;" and "with (nolock)" do.

apart from the fact that it allows you...

The DBA's right about the concurrency issue.  Also, you should really only create a clustered index on a temp table and it should be created before the table...

You're on the right track. Break it up into parts. If possible, get the log to simple recovery during the operation and keep it clear between chunks of processing. Otherwise,...

That's because based on your basic data set, no combination of server and database has a count above one. Put a second instance of Server4 db2 in there and you'll...

I'm sorry. I guess I completely misread that. I tracked down the DOP_FEEDBACK page, and yeah, it's not in Azure. Interesting.

Looking into it some more, the basic batch &...

You can use sys.dm_db_index_usage_stats to see the information on how an index is behaving. I assume you're using that. However, as you find, that info resets for a number of...

In fact, the Microsoft documentation says just the opposite: "All IQP features are available in Azure SQL Managed Instance and Azure SQL Database." You can read about it here.

So,...

And get backups of your databases and test them by restoring them somewhere. Don't just take a backup. Test a restore too. It demonstrates that yes, the backup is good,...

Yep, looks right to me too.

As currently configured, to move an Azure SQL Database to on-premises, unless you're setting up a failover group (which, if you're using Managed Instance can be done), the only choice...

Whatever will make the behavior the same. I can't tell from where I sit.

So, when we talk about comparing performance between something like a UAT system and a production system, we've got to get more specific.

Are you saying, for a given query, against...

For the first execution of a query, you only store a plan stub (basically a hash for the query, plan & settings affecting the plan) instead of a plan.  That's...