Welsh Corgi (12/13/2015)

---

The log was not truncated because records at the beginning of the log are pending replication or Change Data Capture. Ensure the Log Reader Agent or capture job...

This!

2, 3 and 5 (mostly, though it'll be for all tables and...

UDFs are slow. They run once per row. The optimiser cannot cost them properly. The problem is the UDF usage.

Nonclustered indexes are not rebuilt when the clustered index is rebuild.

I'd have recommended Jonathan's book, but doing so seems to be redundant. Chapter 1 and Chapter 6 mainly (I think 6 is blocking)

No.

A copy-only log backup doesn't...

Then you're going to need to create a calendar table that has all months in it (and just columns related to the dates) and query that table LEFT JOIN to...

Customer seems to want a read-write DB where the changes are discarded...

No. Why would there be?

Would you really want an index rebuild that could take hours in the middle of a delete query's execution?

It's the operator which does the actual delete of the rows from the table. Looks like you have a cascading foreign key, so when you delete from the parent table,...

Are there rows for november in your calendar table dbo.Filteredfs_carvelgmrcalendar?

Which will not do a thing in this situation, as the log is full due to an active transaction.

Just based on those queries and no other:

Index ON Article (websiteid, IsActive, startdate, enddate) include (isfeatured ,istopstory ,isbreakingnews ,islatestnews)

Or, as an alternative, if the filter for IsActive is ALWAYS IsActive...

In a properly tuned OLTP database, it should be milliseconds. An analytics system would probably take longer-lasting locks because it's doing larger queries. It's not about normal, it's about whether...

You can try that. It does mean that your invoicing will be serialised and so might run longer and be slower. If that's acceptable, give it a try.