Category: Show Notes

The Recording

The Panelists

• Kevin Feasel
• Mala Mahadevan
• Mike Chrestensen

Notes: Questions and Topics

Data Professional Salary Survey 2026

The first major topic of the night was about a survey Brent Ozar runs annually regarding data professional salaries. The 2026 results are in and I did a bit of poking around with a pivot table. We didn’t do any real in-depth analysis simply because this was my first look at the data. But we did talk a bit about inflation rules of thumb, sample sizes, and a few interesting things in the dataset. I definitely appreciate Brent continuing to put this together and making it available for people to use.

Azure Pricing and Exchange Rates

The next topic covered Azure pricing and exchange rates. Thomas Rushton retold a story (in blog format) regarding some exchange rate volatility in 2022 affecting British Azure customers because everything is ultimately priced in USD and then spot exchange rates determine the conversion factors to other currencies. For currencies that are normally fairly stable, that’s not a huge deal. But in this case, there was a 15% or so difference, simply because a spike happened during the time of month that Azure bills.

Teaching Technical Topics

Our last topic of the night comes from a John Deardurff blog post. John recently wrote about several tips for teaching people about technical topics and we gave our takes on the tips. In general, I fully agree with John. There are a couple of tips that I don’t necessarily observe, but we did talk about when and why it can make sense to do so.

The Recording

The Panelists

• Kevin Feasel
• Mala Mahadevan
• Mike Chrestensen

Notes: Questions and Topics

Days of Data

The first major topic of the night was about a pair of Days of Data. The first one is that Raleigh Day of Data 2026 is live. You can register (either free or a nominal fee for lunch). You can also submit for our call for speakers for the event. Our event is going to take place on Saturday, May 23rd in Durham.

In addition, Day of Data Richmond 2026 is also live. This event will take place on April 11th in Richmond, Virginia. The registration page for this is also open and the call for speakers will be open until January 18th.

Vector Search in SQL Server 2025

Our other major topic was around vector search capabilities in SQL Server 2025. Mala has been investigating this and shared her thoughts so far.

The Recording

The Panelists

• Kevin Feasel
• Mike Chrestensen

Notes: Questions and Topics

SQL Saturday –> Day of Data

The first major topic of the night was an article from Steve Jones on the evolution of SQL Saturday. Steve has officially announced the new name, Day of Data. We talked a bit about this as well as SQL Saturday Raleigh 2026 (or Raleigh Day of Data 2026?). That’s coming up soon.

TriPASS Call for Speakers

In case you missed it, the TriPASS call for speakers for 2026 is currently up. We’re accepting submissions for our three groups: advanced DBA (virtual, 300-400+ level topics), the main meeting (general purpose, with a preference for speakers who can meet us in person), and data science/business intelligence/artificial intelligence (virtual).

Use Cases for PGlite

Mike hit me up before the show and asked for my thoughts on PGlite. I took a look at the website, speculated a bit on good uses cases, and compared it to a pair of other libraries I’m much more familiar with in SQLite and DuckDB.

DAX Lib: an App Store for DAX Functions

Mike also wanted to talk about DAX Lib, jumping off of a recent Curated SQL post. This leads to DAX Lib, a website hosted by the sqlbi team of Marco Russo, Alberto Ferrari, and Kurt Buhler. The idea is to provide people with a repository for user-defined DAX functions that they can use and re-use in Power BI reports. I think it’s a neat idea and it does fit in nicely with sqlbi’s (paid) Tabular Editor 3 product.

Using FORMATMESSAGE

A member of chat asked us to end with a useful SQL tidbit for newbies, and that reminded me of a Louis Davidson blog post on using FORMATMESSAGE. This was built on a bit of yes-and development between Louis and me and I was able to show the audience the final product and how the FORMATMESSAGE() function can make this sort of string formatting considerably easier. At least, so long as you aren’t using date types (or don’t mind converting them to strings first).

The Recording

The Panelists

• Kevin Feasel
• Mike Chrestensen

Notes: Questions and Topics

SQL Server 2025

Our major topic was the release of SQL Server 2025, and this took up pretty much the entire show. We looked at many of the new features in the product, discussed expected adoption curves, and I punched down on SQL Server Integration Services for a while. But in fairness, if they didn’t want me to complain about it, they’d actually have provided meaningful updates to the product this decade.

The Recording

The Panelists

• Kevin Feasel
• Mala Mahadevan
• Mike Chrestensen

Notes: Questions and Topics

The Night before Ignite (and PASS Summit)

We briefly talked about how PASS Summit and Microsoft Ignite were both kicking off, but no news had leaked yet. Since then, we learned about the release of SQL Server 2025.

SQL Server and Automatic Indexing of Foreign Key Constraints

Next up, we talked about SQL Server Management Studio 22 going GA last week. Rob Farley showed it off during the TriPASS meeting, and I spent some time walking through Vlad Drumea’s blog post on configuring SSMS 22.

Along the way, someone in chat asked about running SSMS on MacOS and Parallels. At the time, I said I wasn’t sure but it seemed the answer was no. Since then, I saw this post by Randolph West on how to run SSMS 22 on MacOS Tahoe. It seems that it is possible, though not perfectly.

ETL and ELT Tooling

Mala wanted to get some recommendations on ETL (or ELT) tooling. I walked through a Datacamp article on the topic and we talked a bit about some of the tooling options available. Admittedly, I don’t do as much data movement as I used to, so this was more of a “research and report” type of topic, mixed in with a few observations of the tools that I have used in that space.

The Recording

The Panelists

• Kevin Feasel
• Mike Chrestensen

Notes: Questions and Topics

AWS to Bare Metal

The first item of the episode (after I spent way too long talking about my green screen change) was an article that looked back at the past two years after a migration from AWS back to an on-premises data center. I think the Q&A format was a good one and the author has some interesting answers to questions people had at the time, as well as a look back at the true costs and savings. I’m not surprised that they saved a pretty significant amount of money, especially because it seemed like their primary use case was running Elastic Kubernetes Service rather than many platform-as-a-service offerings.

SQL Server and Automatic Indexing of Foreign Key Constraints

The second topic was this Greg Low blog post (via Brent Ozar’s newsletter), asking if it makes sense to add indexes to foreign key constraints. Greg’s answer is a good one and covers the situations in which it does make sense, as well as situations in which it would not make sense for there to be an index. Greg then offers a hypothetical bit of syntax that would cover automatic creation of an index with a foreign key unless you specify otherwise.

Believing What You See

Greg had another thoughtful blog post that I wanted to cover, this time around images and credulity. Greg points out a real-world event—and then Anders, in chat, lets us know that this kind of thing still happens and an incident occurred not even 6 weeks prior in Charlotte—but people automatically discounted the real-life image because of the deluge of AI generated images out there. I did a bit of riffing on older techniques for image manipulation (airbrushing, image staging, Photoshop) and also talked a bit about how modern phones are beginning to cause real problems as evidence in legal proceedings because of built-in tools that manipulate the image, like taking people out of the background of an image.

Tilting with Tuples

Finally, I gave a shout out to Louis Davidson, who is starting a new podcast. The most important reason I did this was because Louis pronounces “tuple” correctly, and that goes a long way toward an endorsement.

The Recording

The Panelists

• Kevin Feasel
• Mala Mahadevan
• Mike Chrestensen

Notes: Questions and Topics

AWS Outage

The first item of the episode covered an outage in AWS that affected a large number of companies, and even affected the ability of people to log into Twitch for chat. And the cause? DNS. Because it’s always DNS.

Along the way, I ranted about a variety of topics, including Extreme Programming.

Conferences and AI-Generated Submissions

The other topic of the day was the number of AI-generated submissions for conferences. Mala and I shared a few thoughts on the topic as I struggled through the incentives associated with this.

The Recording

The Panelists

• Kevin Feasel
• Mala Mahadevan
• Mike Chrestensen

Notes: Questions and Topics

Reading Parquet Files in SQL Server 2022

Our first came from a reader. The brunt of the question is as follows:

I’ve got a vendor who sends us parquet files. A quick google search suggests that SQL server’s polybase feature allows SQL Server 2022 to directly read from parquet files without to do ELT.

Fortunately, I happened to put together a video on how to read local files in SQL Server with PolyBase and MinIO.

A Checklist for Database Reliability Engineers

Our next major topic was reviewing this blog post from Amy Abel on what Database Reliability Engineers should know. The main takeaway I have from it is that 90% of the contents are simply things that good DBAs should know and do. There are some exceptions to this, particularly around knowledge of things like TerraForm or other Infrastructure as Code techniques, but the majority of this I would lump into “Stuff a good DBA should know how to do.”

The Monty Hall Problem

After that, we covered one of my favorite paradoxical concepts in statistics: the Monty Hall problem. I’ve also created a video on the topic that you’re welcome to watch if you want to dig into the simulation further. This concept is one that really drives home the importance of updating knowledge based on new information, and not applying that new information incorrectly.

DBSCAN in SQL Server

Our final major topic was a fun article around implementing the DBSCAN clustering algorithm in SQL Server. I was legitimately surprised that the performance was fine for ~200 records. Granted, 200 records isn’t a huge number or anything, but I figured it would be so sub-optimal compared to R or Python code that we’d see it hit its limits by that point.

The Recording

The Panelists

• Kevin Feasel
• Mala Mahadevan

Notes: Questions and Topics

SQL Server 2025 RC1

Our first topic was the announcement of SQL Server 2025 RC1. We spent some time talking about time to release between versions of SQL Server with the help of the SQL Server Editions website. We also talked about the likelihood of an RC2 or maybe even RC3 given how quickly Microsoft released RC1 after RC0.

Bad Articles about SQL Have Got to Go

The majority of this episode was dedicated to this Peter Wayner article on 13 reasons why SQL has to go. It conflates SQL the language standard with relational database management systems, but that’s fine. What wasn’t fine was how shallow and poorly-argued the whole thing was. I tried hard not to strawman his arguments, but they were not good. As for replacements, my goal at the end was to say that there exist specific domains in which non-relational data models can make sense, and there can be perfectly valid reasons to have non-relational systems. After all, I have an entire talk dedicated to this concept. Where I take umbrage is the idea that a key-value data store, document database, vector database, search indexing engine, or other technology is better than a relational database 95% of the time, or that some non-SQL language is better than SQL as a data-oriented domain-specific language. And that article didn’t move the needle even a tiny bit.

The Recording

The Panelists

• Kevin Feasel
• Mala Mahadevan

Notes: Questions and Topics

Rest in Peace, Phil Factor

Our first topic was Tony Davis’s sad news that Andrew Clarke, a.k.a., Phil Factor, passed away. Mala and I reminisced on some of our favorite articles of his. Mine, to this date, is still his exquisite story around entity-attribute-value. A pair of Mala’s favorites were around SQL Clone and idempotent SQL processing, and they helped her with her own work. I also pointed out that there’s a free e-book of Phil’s funnier articles, Confessions of an IT Manager. I can highly recommend it, and probably should go back and re-read it now that I’ve been a manager.

Copilot in SSMS 22

We also talked about Erin Stellato’s update on what’s next for Copilot in SSMS. Mala pointed out Brent Ozar’s blog post on the topic from earlier this year, and we also took a look at what it takes to use GitHub Copilot for free in Visual Studio.