Yesterday, I showed the stuff you read the most, but enough data – now, let’s talk about feelings. These were the 11 posts I remember with the most fondness, in date order. I did indeed like some of the ones from yesterday’s list a lot, but I’m going to set those aside.

Erik and Tara are Heading Out (Jan 3) – It doesn’t seem right to have a 2019 retrospective without mentioning how this year started. It wasn’t a fun blog post to write, nor a fun business decision to make. However, looking back as the year closes out, I can say with more confidence that it was the right business decision to make, and I’m glad I did it.

Consultants: want a utility to gather SQL Server data? (Jan 8) – As soon as I decided to stop growing our own consulting work force, I decided to start selling the consulting tools that we’d built. The Consultant Toolkit is now used by hundreds of consultants around the world. That’s kinda awesome.

How fast can a $21,468/mo Azure SQL DB load data? (Feb 13) – spoiler alert: slower than a USB thumb drive.

Can deleting rows make a table…bigger? (Mar 5) – Anytime I can help Michael J. Swart solve a problem, I’m all over it. Plus, this issue pops up again in a new way with SQL Server 2019’s Accelerated Database Recovery, but I’ll leave that one for 2020.

Fixing Performance By Throwing Cloud At It (Apr 23) – This was one of those posts that just seemed obvious to me, but it came up repeatedly with clients, so I figured I’d write it down. I wasn’t proud of that post at the time – it just seemed so obvious – but I heard from a few people afterwards that they’ve integrated the term “throwing cloud at it” into their vocabulary, so that’s kinda cool.

15 Reasons Your Query Was Fast Yesterday, But Slow Today (Apr 18) – Even today, it’s surprisingly hard to get definitive answers from monitoring tools when this question comes up.

Pop Quiz: SQL 2000 Bug or SQL 2017 Bug? (May 1) – I knew this post wouldn’t be timeless when I wrote it, but hot dog, I really enjoyed writing it.

Yep, Developers are Using Unmasked Production Data. (July 5) – I get so frustrated when I hear trainers/presenters/bloggers/idealists talk about how developers should be using purpose-built-from-scratch data sets with no real customer data in ’em. The real world just rarely works that way. Sure, I get that it should work that way, but it’s just not reality today, so I wanted to poll the audience to get some hard numbers for proof.

The 6-Month DBA Training Plan (July 15) – Wow, time flies: I first wrote this 8 years ago as an email autoresponder. I kept getting emails asking me to update some of the content, so in the summer, I figured I’d sit down and just turn the private autoresponder into a public series of blog posts, and also update the autoresponder content while I was at it. I thought it would be quick and easy. I was wrong.

“But NOLOCK Is Okay When The Data Isn’t Changing, Right?” (Aug 5) – Simple demo & video proving why you probably shouldn’t be using NOLOCK.

So, uh, you can, like, draw me in SSMS. (Nov 16) – Because it’s amazing to have someone do something like this: