Ah, I didn't see the second post there, sorry. So, the correlated queries are running faster? Hmmm.... I'd want to see both execution plans to understand why.

Oh wow, it's a muck table[/url]. If you ever get a shot at redesign... this should be public enemy #1.

The first query is the preferred method. Not the second. It'll...

Rows, not records. And who cares. 300,000,000 rows at 2k/row is 585,937.5...

rodjkidd (11/29/2016)

---

Grant Fritchey (11/29/2016)

---

rodjkidd (11/29/2016)

---

Steve Jones - SSC Editor (11/25/2016)

---

Hope people had a good Thanksgiving in the US, or a quiet day elsewhere.

Another day off for me, so I'm enjoying...

joeroshan (11/29/2016)

---

By 'select in select' do you mean sub queries? In that case it is better to use joins that sub queries. You may not be noticing any difference currently...

Be very cautious in what you're doing. From my reading of it, you're creating a view that does a whole series of JOIN operations (and a JOIN is the better...

rodjkidd (11/29/2016)

---

Steve Jones - SSC Editor (11/25/2016)

---

Hope people had a good Thanksgiving in the US, or a quiet day elsewhere.

Another day off for me, so I'm enjoying it after a...

Thinking about it, make sure you have updated your SSMS version as well.

I know there's documentation saying that backup to URL is supported in 2012 as of SP2, but it's...

Currently on the MSDN documentation on Backup to URL, it only shows support in 2014 & 2016. There was an app to do backup to Azure for 2008 & up,...

Surely the denizens of...

I don't see any mention of cost threshold for parallelism? What's that set to?

Also, in addition to the cardinality estimator, all sorts of changes have been made to the optimizer....

The biggest issue with upgrading to 2014 is the new cardinality estimator. Overall, it helps, but there are edge cases, especially with queries that were poorly written to begin with,...

Well done to all Steve. That's a hell of an accomplishment.

If you're migrating your data into a new configuration, I wouldn't do it in a single set as you're describing. Instead, break it apart and move only 100k rows at...

Assuming you're outputting to a file, you can just open the file and you'll be immediately viewing the live data. That may not be a lot easier, but it works.