Press "Enter" to skip to content

Author: Kevin Feasel

Azure API Management in front of Databricks and OpenAI

Published 2025-08-06 by Kevin Feasel

Drew Furgiuele has a follow-up:

A few months ago, I wrote a blog post about using Azure API Management with Databricks Model Serving endpoints. It struck a chord with a lot of people using Databricks on Azure specifically, because more and more people and organizations are trying their damndest to wrangle all the APIs they use and/or deploy themselves. Recently, I got an email from someone who read it and asked a really good question:

Click through for that question, as well as Drew’s answer.

Leave a Comment

Microsoft Fabric Data Warehouse July 2025 Recap

Published 2025-08-06 by Kevin Feasel

Charles Webb lays out some updates:

Welcome to What’s New in Fabric Warehouse, where we’ll spotlight our work improving quality, delivering major performance enhancements, boosting developer productivity, and our continuous investments in security. Whether you’re migrating from Synapse, optimizing your workloads, writing SQL in VS Code, or exploring new APIs, this roundup has something for every data professional. With quality and experience at the forefront, we’ve summarized and highlighted key improvements we think you’ll love, organized into three sections:

  1. What’s New
  2. Docs Updates
  3. Roadmap Updates

Read on for that update.

Leave a Comment

Data Dictionaries in Power BI

Published 2025-08-06 by Kevin Feasel

Ben Richardson builds a dictionary:

Have you ever opened a Power BI report and felt overwhelmed by all the columns, measures, and tables?

It can feel like a guessing game trying to figure out what each field represents.

A well-built data dictionary eliminates that confusion, giving you clarity and confidence when exploring reports.

This is one of those bits of documentation that can be incredibly useful but people rarely keep it up to date.

Leave a Comment

Thoughts on Views

Published 2025-08-06 by Kevin Feasel

Joe Celko shares some thoughts on views, as well as recursive common table expressions:

VIEWs are an undervalued and underused feature in SQL. They basically consist of a query that has been given a name, and a parameter list, so can be used like an inline macro. Technically, you’re supposed to think of it as a virtual table. 

The idea of an inline macro or inline function goes way back to the first versions of Fortran and later BASIC. Depending on the product, the syntax might look like this: FN DOUBLE (N) = (N + N). You had to give the name of the Macro, a simple optional parameter list, assignment operator, usually an =, and the text of the computation. It was often important to put parentheses around the body of the macro, or to have a compiler that would do this for you. 

Read on for the article and a platform-agnostic coverage of views. My problem with views is that developers try to use them to cover a lot of ruin and then they ask why such a simple SELECT * FROM Something query is so slow.

Leave a Comment

ACE Drivers and Linked Servers

Published 2025-08-06 by Kevin Feasel

Sean Gallardy has a public service announcement:

There’s been a resurgence of people pointing out dumps occurring in SQL Server when using linked servers with the ACE drivers. It’s been on the MCM email list, forums, SQL Server Feedback site, everywhere, and it’s basically the same response every time… ACE drivers were not made to be used as linked server drivers in SQL Server.

To be fair, how could the company that produces Access and Excel possibly be in contact with the company that produces SQL Server and create a driver that works well?

I have used the ACE drivers for PolyBase, though that was for fairly light-duty work and thus I haven’t seen any dumps. But I guess if you want a higher-quality driver, go with someone like CData.

Leave a Comment

Reasons Regression Models Under-Perform

Published 2025-08-05 by Kevin Feasel

Ivan Palomares Carrascosa has a list:

In regression models, failure occurs when the model produces inaccurate predictions — that is, when error metrics like MAE or RMSE are high — or when the model, once deployed, fails to generalize well to new data that differs from the examples it was trained or tested on. While model failure typically shows up in one or both of these forms, the root causes can be more diverse and subtle.

This article explores some common reasons why regression models may underperform and outlines how to detect these issues. It is also accompanied by practical code excerpts using XGBoost — a robust and highly tunable ensemble-based regression model. Despite its popularity and power, XGBoost can also fail if not trained or evaluated properly!

These are high-level reasons but they’re good to keep in mind.

Leave a Comment

Updates to Extensions in C# 14

Published 2025-08-05 by Kevin Feasel

The word miners in the NDepend blogging mines clue us in on an update to the C# language:

A few remarks:

  • If the extension is generic, the type parameter (if any) is specified immediately after the extension keyword.
  • The receiver ReadOnlySpan<T> span applies to one or more extension methods (or members) declared within the same extension block. This logical grouping is a core advantage of the new syntax, improving clarity and structure when extending a type.
  • As far as the C# compiler is concerned, the two methods Truncate() are strictly equivalent. Therefore, one must be commented out for the program to compile.
  • Notice the call to AsSpan() in the first line. Extension methods require the receiver type to match exactly ReadOnlySpan<T>, hence it doesn’t apply to string unless you explicitly cast. This is not specific to the new syntax but worth mentioning.

Read on for examples of how extension methods and extension members will open up opportunities to improve existing C# code.

Leave a Comment

Regular Expression-Based String Splitting in SQL Server 2025

Published 2025-08-05 by Kevin Feasel

Aaron Bertrand splits a string:

SQL Server users have been asking for native regular expression support for over two decades. There are third-party Common Language Runtime (CLR) modules that offer this functionality, but these can be complicated to install and simply aren’t possible in some environments. I want to split a string using a regular expression instead of a static string. Will that be possible in SQL Server 2025, without CLR?

Must not rant about CLR. Must not rant about CLR. Must not rant about CLR. (By the way, if you ever catch me in person, get me going about how CLR got the short end of the stick and how the ‘modern’ forms of the Common Language Runtime in SQL Server are not great.)

Aaron tries out a function built into SQL Server that allows you to split strings into result sets using a regular expression to perform the splitting, and shows off some of the more complicated scenarios that this can solve over a normal STRING_SPLIT() function call.

Leave a Comment

Using a Child Pipeline Variable in a Parent Pipeline in Fabric Data Factory

Published 2025-08-05 by Kevin Feasel

Justin Bird passes back some information:

I answered a question on the Fabric community on return variables recently and thought I would expand upon it in a blog post. The question was how to use a variable derived in a child pipeline downstream in the parent pipeline. The person was specifically deriving a json object and wanted to iterate on the values in the parent pipeline.

Click through for the solution.

Leave a Comment

Performance Optimizing PostgreSQL for RTABench Q0

Published 2025-08-05 by Kevin Feasel

Andrei Lepikhov gets under the hood:

I wanted to explore whether Postgres could be improved by thoroughly utilising all available tools, and for this, I chose the RTABench benchmark. RTABench is a relatively recent benchmark that is described as being close to real-world scenarios and highly selective. One of its advantages is that the queries include expressions involving the JSONB type, which can be challenging to process. Additionally, the Postgres results on RTABench have not been awe-inspiring.

Ultimately, I decided to review all of the benchmark queries, and fortunately, there aren’t many, to identify possible optimisations. However, already on the zero query, there were enough nuances that it was worth taking it out into a separate discussion.

Click through for a dive into this particular query and what Andrei did and some of the lessons you can draw from it.

Leave a Comment