ZenDada (6/25/2010)

---

As usual, Pinal Dave has the 411!

SQLJocky (6/22/2010)

---

Sure did. It works fine as long as it was a controlled failover. If it was an automatic failover or forced manual failover, it holds the transactions...

Did you test updating the "subscriber" when mirror server took over?

I am not 100% but I think it will fail!

Not sure if it is just me but I had a completely different expectation of what an "Expert System" would be 🙂

Nice story though.

Check permissions of snapshot agent account on that Folder!

you will be much better off by composing those statements on the "client" instead of on the "server"

This is great!

I really don't like M$ attitude about trace Flags. Why do we have to go through so much trouble to know (and compile) these things?

johan.lindell (5/24/2010)

---

Good comparison, David, thanks.

From a performance perspective, using NOT IN is "always" slower than using EXISTS. If you try running the below queries

-- INTERSECT

SELECT CustomerID

FROM Sales.Customer

WHERE TerritoryID=10

AND NOT EXISTS...

Also what initialization mode did you use ?

Also you probably want to "Page" that resultset 🙂

looks like your log drive and your tempdb drive are on C

I can only assume that you have a design problem with some of your tables.

You probably chose the wrong replication method. Why can't you do transactional instead?

I don't know what is all this fuzz about ... The results are going to be very server-dependent you don't know what other activities are happening at the same time...

you are probably trying to substitute a "similar" character. Please verify what is the correct ASCII code or UNICODE for that character.

By the way your query works!

declare @t1 table ...