Jeff Moden (12/5/2015)

---

I'm thinking that the logical to physical CPU ratio on that box is the real...

Fandy08 (12/7/2015)

---

Yes, I checked, a few column statistics are older than those of indexes, do you mean Orlando that performance slow down can be caused by these differences on update...

You don't have to bet. From the OP:

SELECT FeedName ,

'Missing FileTableDB Record - External...

Because it's perfectly legit SQL, syntatically valid. The fact that it's not what...

And it will only do anything if the cause of the full log really is replication. Plus, it'll break transactional replication and require that the subscriptions be reinitialised and it'll...

What's the lock escalation set to for the partitioned indexes?

What's the logic there? What is it that defines that the first row listed there must come first?

Updating every column's going to make update triggers more complex, as you now can't use IF UPDATE(... or similar, as every column will return true.

Not necessarily. RowNumber can do grouping, and so return one row per partition. Top 1 is just top 1.

Yes, that would have updated all index stats. Probably resulted in the optimiser realising there was more data than it had previously thought (due to stale stats) and so changing...

Just on this point.

My 2-year old laptop has more memory than that, you're...

Scalar functions are slow. What probably happened is that the execution plan changed and as a result the function was being run many more times.

Don't 'prove it's a san issue'. Go sit with the SAN admins and debug this together. Pull the IO latency stats from the server and go and sit with them...

KILL <session ID>

However do that and it will roll back, and roll back usually takes longer than the operation had up to that point. So kill an index rebuild that's...

Tablediff works. Custom scripts. Redgate's SQLDataCompare.

Better to test comprehensively before the upgrade and not end up in a state where you have to revert. Identify all the breaking changes before...