The full, the latest diff, all the logs since that diff. The fact that earlier diffs are corrupt is irrelevant.

Yes, but that's fine. That's why I said ignore the execution times of the first run (the one that loads the data into memory). After that, the times are consistent...

Don't ever append backups to the same file. If you do and there's damage to the header, you can lose all of the backups. Backups should go to individual files,...

Count(*) can easily be faster than select *. Firstly SQL will use the smallest index on that table (assuming no where clause). Second, less data to transmit.

DropCleanBuffers has nothing to do with the plan. It clears the data cache. You don't need to run it for performance testing, just run queries twice or more and ignore...

It's worth noting that rebuilds aren't themselves replicated.

The actual query was a little (lot) more complex...

You can rebuild it.

What you can't do is drop the primary key, not because it'll break replication, but because you'll get an error if you try.

WHERE Coalesce(NotNullableColumn,-1) = 0

and CategoryID = @CategoryID

Group By CategoryID

......

So a sampled update. That may be fine, or it may be terrible. Can't tell. Take a look at your poorly performing queries...

Jack Corbett (7/9/2014)

---

1. Scaling - ORM tools still aren't very good at complex SQL (multiple joins, outer joins, etc...) so you need to evaluate every query that EF...

I wouldn't say that.

Fragmentation is not that massive of a concern, but rebuilding indexes also removes the half-full pages that you tend to get as a result of page splits....

Board games on some Saturdays at the local geek hangout. I have Catan, Star Flux, Avalon and Forbidden Island. The club usually has a large pile of demo games. Computer...

What recovery model is the DB in?

They would reference a linked server name, which does not have to correspond with the actual server name. When you create the linked server on the new installation, keep the...