• Benni De Jagere

Deep Dive into Microsoft Fabric Capacities: Fundamentals, Concepts, and Advanced Strategies

Microsoft Fabric capacities are powerful, flexible, and at times deeply intricate.

In this advanced session, we’ll unpack the full lifecycle of capacity behavior, starting with the fundamentals: how bursting delivers peak performance beyond SKU limits, how smoothing spreads compute usage across timepoints, and how throttling protects shared resources under sustained load.

Building on these core mechanisms, we explore advanced capacity operations including pausing and resuming workloads, scaling capacities up or down, and making effective use of reservations for predictable cost control.

We’ll then shift to data‑driven optimization strategies, using real usage patterns to diagnose overloads, interpret smoothing debt, and identify when performance bottlenecks require tuning or scaling.

You’ll learn how to leverage Fabric’s built‑in metrics, including insights from the Capacity Metrics App and platform telemetry, to develop a proactive, evidence‑driven monitoring approach.

By the end of the session, you'll be equipped to operate Fabric capacities with confidence—balancing performance, governance, and cost in complex enterprise environments.

• Ásgeir Gunnarsson

From Chaos to Control: Orchestrating Lakehouse Workloads in Microsoft Fabric

You have built a lakehouse in Microsoft Fabric. The notebooks are in place, data is flowing through the layers, and transformations are working. Now comes the hard question: how do you orchestrate it all, reliably, scalably, and in a way that fits real-world workloads?

Should you use Data Factory pipelines, notebook-driven orchestration, materialized views, or bring in a third-party orchestration tool? Each option has clear strengths, but also trade-offs that can become painful in production if chosen incorrectly.

In this session, we take a deep, practical dive into the orchestration options available for Fabric lakehouses. We will break down how each approach actually works, where it excels, where it falls short, and, most importantly, when you should use one over the others. Through concrete scenarios and production-proven examples, we will compare orchestration patterns for batch processing, dependencies, retries, monitoring, and operational complexity.

This is not a theoretical discussion. The session is grounded in real implementations used in production Fabric environments, including the challenges, pitfalls, and lessons learned along the way.

After attending this session, you will leave with a clear decision framework for choosing the right orchestration strategy for your Fabric lakehouse, confidently and deliberately, rather than by trial and error.

Slide deck etc

From Chaos to Control Orchestrating Lakehouse Workloads in Microsoft Fabric_Asgeir Gunnarsson.zip

• Thomas Kejser

Ask the Database Doctor anything

We will start off with a high speed tour of how query optimisers work and how to compare databases with each other. This will be a 1 slide per minute type of session - just to get your warmed up!

Then, we open the floor for Q/A - ask me anything you like!

I've worked with databases for over 35 years and hacked PostgreSQL, SQL Server, Yellowbrick and now FloeDB. This includes writing code (I wrote an execution engine), spec'ing and validating entire subsystems, blogging, tuning benchmarks and hand crafting more SQL query plans than I can remember.

Having spent my life this way I hope to have some experiences to share. I will do my best to answer your questions and whiteboard ideas, explanations and solutions.

• Just Blindbæk
• Greg Baldini

Breaking the Limits: Advanced Optimization of Semantic Models in Microsoft Fabric

As semantic models grow in size and complexity, performance, memory usage, and refresh times quickly become the biggest challenges in Microsoft Fabric.

In this advanced, hands-on workshop, we go beyond best practices and dive deep into how the VertiPaq engine behaves under real-world pressure. You will learn how to diagnose bottlenecks, reduce model size, and optimize performance using proven techniques from large-scale enterprise solutions.

We will explore topics such as column cardinality, encoding strategies, memory consumption patterns, and how different storage modes—including Direct Lake—impact performance and scalability.

You will also learn how to design efficient refresh strategies, avoid common pitfalls that lead to paging and memory pressure, and understand how query patterns affect performance.

Through live demonstrations and interactive optimization challenges, we will diagnose intentionally inefficient models and transform them into production-ready solutions.

This workshop is based on real-world experience optimizing semantic models in Microsoft Fabric and Power BI at scale.

If you are working with large or complex models and want to push performance to the next level, this session is for you.

Slide deck etc

Advanced Optimization of Semantic Models in Microsoft Fabric.pdf

• Marc Lelijveld

From Source to a Working Agent in 3 Hours: Building an AI Ready Lakehouse in Microsoft Fabric

In the era of AI, a “working agent” is only as good as the data foundation underneath it. In this 3 hour deep dive, we’ll go from raw source data to an AI ready Microsoft Fabric Lakehouse and finish with a data agent that can answer meaningful business questions—reliably and repeatably.
We’ll build a well architected Lakehouse using a medallion approach (Landing Zone → Bronze → Silver → Gold) and show how those layers translate into robust ingestion, transformation, historization, and curated consumption for analytics and agents. From there, we’ll connect a semantic/consumption layer and discuss why a clean model and clear metadata are critical for agent accuracy and adoption.
Finally, we’ll bring it together with Fabric Data Agents concepts, tools, context, configuration and structured data access patterns, so you understand what it takes to move beyond demos and build something you can actually use in real projects.

Slide deck etc

2026-10-09_From source to working agent.pdf

• Simon Whiteley

Data Engineering - Now & the Future

We start with the job as it actually is. Design, analysis, coding, more coding, testing, glorious documentation. AI turns up differently at every one of those stages — genuinely good at some, oversold at others, quietly useless at one or two. We will work through them, look at what people are really doing rather than what the demos promise, and show where it holds up.

Then the harder half. What changes when the tools stop assisting and start operating? They start running the platform, owning the pipeline, fixing things before anyone has noticed they broke. This isn't a dream, there are already tools being released that do exactly this. When they are the pilot rather than the copilot, where does that leave the humble engineer? I have opinions. I do not have the answer, and anyone who tells you they do is selling something.

The format is the same throughout: I'll introduce a topic, share what I know, demonstrate it where there is something worth showing, then open the floor and argue about it. Roughly half the afternoon is prepared, the other half is yours. Bring the questions you have been chewing on: about the tools you are using, and about the industry you are in, and the one you are about to be about to be in.

• Cláudio Silva

VECTOR in SQL Server 2025: When, Why, and How to Use It

SQL Server 2025 introduces a new data type: VECTOR.

In this session, we will explore what exactly the VECTOR data type is and the problems it helps solve.

I have been doing some proof-of-concept with clients and want to share practical tips, challenges encountered and where these new functionalities fit your scenarios.

Some of the questions we will be answering during this session include:
- What capabilities does it bring that were difficult or inefficient before?
- How does it impact table design?
- Being a new data type, can VECTOR columns be indexed, and how/why might that change the results?

We will also see it in action, how it powers semantic searches by using Similarity search and some more advanced Hybrid searches scenarios, where we combine it with the traditional relational filtering.

Slide deck etc

Data Saturday Holland - VECTOR in SQL Server 2025.zip

• Ásgeir Gunnarsson

Find the Spark as a SQL data warehouse developer

You are a data warehouse developer that´s been using SQL Server on-premise or in Azure to build your data warehouses. You are comfortable in writing SQL scripts, stored procedures and creating views to fuel your data warehouse. You have heard about the new kid in town Spark but you have not yet take the plunge and are wondering if you have to or if you should.

If the above is something like you then this is the session for you. The session is built largely on my own experience. I had been doing data warehouse in SQL for years and I had decided Spark was nothing for me. I felt there was enough to do in SQL so there was no reason to add Spark to the tools I was using. It seemed like a lot of learning and when you are an old dog that can be difficult to do 😊.

The reality is that Spark offers some very cool features and it´s not that different to how you have been doing things. Actually the concept of how to do a data warehouse do not change that much even though it´s built in a data lake and is called a data lakehouse. Yes the technology is different and the language is a lot or slightly different depending on what you choose but transitioning was lot easier than I thought.

We will start by comparing SQL data warehouse and Spark lakehouse. What is common and what is different. Then we will look at a demo on how you could do things in a Spark lakehouse all the time comparing it to how we are used to doing it in SQL data warehouse. We will then move onto understanding how to acquire the right knowledge to get started with Spark as well as how to try it out without breaking the bank.

After attending this session you will be better able to understand if Spark lakehouses are for you and how you can figure out how to learn the required skills. You will also know how to get started for free or with a very low cost.

Slide deck etc

FindingTheSparkAsSQLWarehouseDeveloperAsgeirGunnarsson.pdf

• Artur Tyvaert
• Parcifal Van Gucht

How to expose business users to AI applications in a controlled environment

Claude, chatGPT, Gemini, skills, tools, MCPs... These are all buzzwords tossed around way to frequently lately. New technologies are emerging at an incredible pace and people are reacting in different ways...

• They keep up-to-date with all the latest hot trends although this is a big challenge
• They nod along and pretend that they understand everything while they have no clue what is going on.
• They start to worry they are getting left behind.
None of these options are a long-term solution for business users and as data professionals it is our responsibility to bridge this gap. Our goal should be to safely expose business users to these new technologies so their valuable business logic can be paired with the power of AI.
But how can we achieve this? In this session, we will explore how Databricks brings all the necessary features together into a single, unified, controlled experience for business users. Instead of downloading Visual Studio Code, users can go to their Databricks App where they can ask question about their data, update a sharePoint file or trigger a tool in a managed environment without have to understand what an MCP or tool is. For business users, this will feel like a much more natural experience than invoking an tool from a terminal.
In this session we will go over how this can become a reality and what limitations it has. Why should you use Databricks Apps over the Genie interface provided by Databricks? How do you set up a Databricks App and expose it to end users? What limitations does this have? How can business people use conversational BI to extract value from their data and trigger actions based on this information? How do you manage access to all these different options and how do we monitor user behaviour? What role can Lakebase play in this architecture? And at the end of it all, is this set-up worth the price and the effort?
Do these questions intrigue you and do you want to know how all these concepts can be brought together into one unified platform? Then be sure to check out our session!

• Danny de Haan

Database Development with a Security-First Mentality

In today’s increasingly digital world, ensuring the security of data is paramount. This session, will dive into the essential practices and principles of secure SQL Server database development. Attendees will learn how to adopt a proactive approach to database security from the initial stages of development, incorporating best practices to safeguard sensitive information against unauthorized access, data breaches, and malicious attacks.

The session will explore critical security features within SQL Server, such as encryption, authentication, role-based access control, and auditing. Furthermore, we will address the importance of regular security assessments and continuous monitoring to stay ahead of evolving threats.

By the end of the presentation, developers will have the knowledge and tools to integrate security seamlessly into the database development lifecycle, ensuring their SQL Server databases are resilient and compliant with industry standards and regulations.

Slide deck etc

Database Development with a Security-First Mentality - DSH.pdf

• Kasper de Jonge

Security and Data Governance in Microsoft Fabric: A Practical Implementation

Microsoft Fabric brings data, analytics, AI, and business users together on one SaaS platform. This session explains how to secure that environment using layered controls and practical implementation guidance. We will connect identity, networking, workspace and item permissions, OneLake security, governance, and monitoring into one security model.

Areas covered
- Zero Trust and Conditional Access
- Network isolation and Outbound Access Protection
- Workspace, item, and OneLake security
- Encryption, sensitivity labels, and DLP
- Audit, monitoring, and investigation
- Shared responsibilities across Fabric, identity, governance, and security teams

• Jesse Gorter

The Limitless Data Platform - The Rise of Kubernetes

For years, modern data platforms have increasingly moved towards managed services such as Microsoft Fabric, Databricks and Snowflake. These platforms are powerful, easy to adopt and come with many capabilities out of the box.

At the same time, many organizations are running into new questions around platform costs, vendor lock-in, architectural flexibility and operational control. As a result, more and more data professionals are asking themselves:

Should Kubernetes become the foundation of a modern data platform?

In this session, we explore how to build a complete lakehouse architecture on top of Kubernetes using open technologies such as Delta Lake, DuckDB, object storage and containerized workloads.

Rather than focusing on theory, we will walk through the architecture of a real platform built from scratch and discuss the design decisions, challenges and trade-offs encountered along the way.

Using StatsBomb football data as a practical example, we will demonstrate how data flows through the platform from ingestion to analytics and reporting.

Topics covered include:
• Why organizations are looking beyond traditional managed data platforms
• The role of Kubernetes in modern data architectures
• Building a layered lakehouse architecture using Delta Lake
• Using DuckDB for analytics
• Designing ingestion, integrated and serving layers
• Lessons learned when building and operating a platform yourself

By the end of this session, attendees will understand when Kubernetes is a realistic option for a data platform, which architectural patterns work well in practice, and what it takes to combine open technologies into a scalable and modern lakehouse platform.
Including live demos, real code and real football data.

Slide deck etc

The_Limitless_Data_Platform_Data_Saturday_2026.pdf

• Mladen Prajdic

SQL Server 2025 Optimized Locking

Introduced in SQL Server 2025, optimized locking improves transaction locking mechanism
to reduce lock blocking and lock memory consumption for concurrent transactions.
It's composed of two primary components: transaction ID (TID) locking and lock after qualification (LAQ).
In this session we'll look into how this works and how we can use it.

• Lars Andersen

Are your semantic models ready for Copilot?

Most Power BI semantic models were built before Copilot existed. They still deliver value and valuable insights. Imagine that your manager tell you that you should enable Copilot for Power BI and make consumers more productive - what do you do?
Simply turning on Copilot for existing models won’t guarantee results. Why? Because Copilot relies on well-structured, optimized semantic models to deliver meaningful insights and efficiencies.
In this session, you’ll learn:
• Why preparation matters: Understand the limitations of “as-is” models and what Copilot needs to shine.
• Best practices for optimization: Practical steps to make your semantic models Copilot-ready.
• Ways to consume models with Copilot and Agents: Explore different integration scenarios and user experiences.

Whether you’re a BI professional, data modeler, or decision-maker, this session will equip you with actionable strategies to maximize Copilot’s impact on your organization. This session will include live demos, including optimizing models for Copilot and consuming Copilot.

Slide deck etc

Are your semantic models ready for Copilot.pdf

• Linda Torrång

Microsoft Fabric Performance Pitfalls and How to Avoid Them

Performance issues in Microsoft Fabric rarely have a single cause — they’re the result of small decisions adding up along your entire data journey. In this session, we’ll follow that journey from start to finish, exploring common pitfalls and practical ways to avoid them.
We’ll look at key performance considerations in every stage of your architecture — from ingestion to storage, through transformation, and all the way to reporting in Power BI.
Whether it’s designing for incremental refresh, managing partitions and lifecycle in your lakehouse, optimizing notebooks and pipelines, or making smarter choices in your reports — this session will give you a high-level map of where performance problems typically sneak in, along with hands-on tips for staying in the fast lane.
Expect a mix of architecture guidance, real-world lessons learned, and actionable advice you can take back to your own Fabric environment.

Slide deck etc

Fabric_Performance_Pitfalls_DataSatHol2026.pdf

• Traci Sewell

Talking about mental health - For you and for them

Looking after more than just 1's and 0's is important for us.

Everyone experiences, or is adjacent to people who experience, challenges with mental health. In both the workplace and personal life, it can feel very difficult to talk with people about these challenges.

We will outline potential behaviours or changes in behaviour that may indicate mental health issues to help you to better identify them in yourself and in others.

We will also provide you with a small set of tools to help navigate these uncomfortable conversations, making them easier for everyone involved and helping you to be able to better help those around you.

You will leave this session with some tools to be able to help yourself identify and deal with these challenges and a better understanding and knowledge of how to be most supportive for your family, friends and colleagues who are facing them.

You will also know that you are unlikely to make it worse !

• Yvette Lovas

Design Thinking in BI Product Lifecycle

There is no such thing like two exactly the same BI report or project. The requirements are always evolving, so does the business. Due to the knowledge and data democratisation trends, the stakeholder group is even more heterogenous now than ever before.

To address this problem, different approaches can be followed during product delivery, and Design Thinking method could easily establish a common ground where the users, the designers and the developers can collaborate to create data products beyond expectations.

During my session I would like to present:
• how Design Thinkin approach can be applied,
• when the wireframing method can be useful,
• where to start gathering feedbacks,
• which questions needs to be asked and interpreted,
• how establishing visual standards can support the process,
• how delivering simple products may have significant effects on data literacy,
• what are the different approaches to effectively train & onboard your users,
• how user adoption can/should be tracked.

Slide deck etc

DesignThinkinginBILifeCycle_YvetteLovas.pdf

• John Martin

Modernizing workloads from SQL Server to PostgreSQL with Flexible Server for PostgreSQL

There are many reasons to look at using PostgreSQL for your database systems, but what if you're already using SQL Server?

In this session we will walk through the process needed to migrate a SQL Server database to PostgreSQL running on Azure Database for PostgreSQL Flexible Server. Then take a practical look at how to put it into practice converting the database code, moving data, and automating deployments with Azure Services and free community tools.

You'll walk away with the knowledge needed to start moving workloads from SQL Server to PostgreSQL on Azure.

Slide deck etc

SQLServertoPostgreSQL.pdf

• Sander Stad

DevOops: What Not to Do in DevOps

DevOps isn’t just about tools, it's about culture, collaboration, and discipline. But what happens when it goes wrong?
This session explores common DevOps anti-patterns.
We’ll cover issues like ignoring databases in CI/CD, manual deployments, broken pipelines, and teams working in silos.
Learn from real-world failures. Discover ways to turn your DevOps dream into success, not a nightmare. Database DevOps mistakes included, because yes, your SQL scripts matter too.

• Ashley Warren

Becoming a Geni(e)us: How to get the most out of Databricks Genie Agents

Databricks genie agents are an excellent tool for helping your business engage with your data. Having the ability to chat with your data, enables business users to get insight faster and closer to the source. But how can we ensure that Genie is accurately answering our user's questions and what do we need to consider before we rollout it out to in our organisations?

In this talk, we'll look at the best practices for running and maintaining a Genie agent. We'll setup a genie agent and discuss what you need to do upfront to ensure that they work for your users. Then we'll walk through the ongoing monitoring and maintenance of a Genie agent & what you can do when users aren't getting the responses they expect. Finally we'll touch on the wider governance aspects of operating genie spaces, including when you should and shouldn't consider Genie agents for your use case

Are you ready to optimise your genie agents?

• Just Blindbæk
• Greg Baldini

Choosing the Right Storage Mode: Import, Direct Lake, DirectQuery & Hybrid

Fabric introduces powerful storage options - but also real confusion. With Import, Direct Lake, DirectQuery, Composite, and Hybrid tables all on the table, it's rarely obvious which mode to choose for performance, freshness, governance, or cost.

This session cuts to the chase with a clear, practical decision framework. We walk through how each mode behaves internally, the trade-offs between memory and compute, common anti-patterns, and the scenarios each is best suited for. You'll leave knowing how to design solutions that balance fast performance, low refresh load, and a seamless end-user experience.

Slide deck etc

Choosing the Right Storage Mode.pdf

• Rich Benner

Using AI for Performance Tuning in SQL 2025

When performance problems strike, DBAs and developers are under pressure to find answers quickly. In SQL Server 2025 AI tools can now assist by analysing code, execution plans, and performance metrics in seconds.

This session demonstrates how to use AI to troubleshoot common SQL Server performance issues, including inefficient queries, missing or misused indexes, and sub-optimal query patterns. We’ll walk through real examples, showing how AI recommendations compare to traditional tuning techniques and how to combine both approaches for better results.

You’ll gain practical guidance on using AI responsibly, understanding its output, and avoiding common pitfalls while tuning mission-critical workloads.

• Robert Hartskeerl

My database will be hacked. But I don't care.

My database will be hacked. But I don't care. It is a bold statement. But reality is that there is a good chance your database or application will be hacked at some point. Data will be stolen, or held hostage. But by far the biggest problem is the impact it will have on the people that trust that their data is safe in your hands.

Still a lot of people seem to be not aware of this risk. Or think that it doesn't apply to them because they are a special kind of company that will never be exposed. Truth is that every company can be a target at any moment. So, with that in consideration what are the risks you see?

This purpose of this session is to get a discussion going at you company by identifying the current risks and if you can mitigate them. The ultimate goal would be that your company gets hacked, someone gets access to the database but no usable data was stolen.

• Marjolein Opsteegh

Deep dive into Power BI theming: from modern defaults to custom JSON

Power BI theming has changed a lot. With Modern Visual Defaults (Fluent 2), many things we used to fix with JSON are now part of the standard experience: cleaner visuals, consistent spacing, subtitles, rounded corners, and built‑in style presets per visual.

In this demo‑driven session, we take a fresh look at Power BI theming—starting from Microsoft’s updated base theme, and then moving beyond it where customization is still needed.
We’ll move up and down the theming stack:

✔️ Explore the new default base theme and modern visual defaults to understand what Power BI now gives you out of the box
✔️ Look at style presets, how they affect visuals, and how you can adjust or override them
✔️ Customize the theme via View > Customize current theme to see how structural settings flow through the report
✔️ Export the theme and fine‑tune it in JSON for advanced or unsupported properties
✔️ Clean up and adapt older themes to work with the new defaults instead of against them

The session fully incorporates the Modern Visual Defaults & theme improvements preview, showing how Microsoft’s new direction changes the way we should approach theming and report design.

This session is aimed at experienced report builders who want consistent, scalable design without repetitive formatting.

Everything is shown live in Power BI, no slides, with a downloadable theme file to take home and adapt.

Slide deck etc

JSON - DSH.pdf

• Juliana Smith

POURing Accessibility into Power BI: Practical Steps for Inclusive Analytics

Many organisations focus on accessibility for websites, but overlook Power BI, even though reports used for decisions, risk, and performance are digital products too.

If people can’t access them, they’re excluded. My perspective is shaped by experience: after a work injury caused nerve damage, a simple hardware change transformed my ability to work and reframed accessibility as essential, not optional.

This session introduces the Web Content Accessibility Guidelines (WCAG), which define global standards for digital accessibility, and explains how their POUR principles (Perceivable, Operable, Understandable and Robust), translate directly into practical Power BI design choices. We also explore how AI/DAX/UDF can generate meaningful alternative text for visuals, improving accuracy and reducing manual effort.

Attendees will leave with clear, repeatable methods to make Power BI reports more inclusive, usable, and compliant by design.

Slide deck etc

POURing Accessibility into Power BI - Juliana Smith.pdf

• Enrico van de Laar

From Convenience to Control: Breaking Free from Big Tech Data Platforms

Many modern data platform infrastructures - like those you can build in Microsoft Azure or directly use like Microsoft Fabric - offer convenience, require less technical know-how, and provide direct integrations into other services. At the same time, they also create large dependencies on their platforms that are deeply rooted and difficult to break.

However, with the current geopolitical climate and tensions between the US and Europe, many European organisations are questioning what it really means to rely heavily on US-based technology and are looking for ways to break away from those dependencies and move towards more digitally sovereign solutions.

In this session we are focusing on a practical path forward: how to migrate from Microsoft Azure and Fabric towards a sovereign data platform that is fully in control of your organisation and at the same time complies with all relevant European legislation.

We will map existing Microsoft components to open-source or European alternatives, design an architecture that is open, modular, controllable, and vendor-independent, and of course, discuss the trade-offs involved in moving away from Big Tech data platforms.

Instead of a theoretical discussion, this talk dives into real-world migration strategies and lessons learned from helping organisations move towards a data platform infrastructure without dependencies and fully in their own control.

• Sebastian Steinau

Getting Started with Spark Internals

Spark is everywhere today, from Microsoft Fabric to Databricks to Amazon emr and others. With AI-powered code generation, ETL pipelines are being created faster than ever before. But faster code does not automatically mean good code.
Without a fundamental understanding of Spark, resources are quickly wasted, runtimes increase, and capacity costs rise significantly. When everything is slow, you just scale up and lose even more money.

In this session, we will take a look behind the scenes of Apache Spark:
How is Spark built internally? How are workloads distributed? And which coding decisions have a direct impact on performance and costs?

We will cover practical topics such as:

- How Spark processes data and distributes workloads
- What Lazy Evaluation means and why actions trigger the actual execution
- What a shuffle is, why it is so expensive, and what causes it
- When caching and persistence are useful and when they become harmful
- Which join strategies exist and when each one should be used
- How to correctly read and interpret the Spark UI
- How to use this knowledge to identify and optimize bottlenecks effectively

The goal of this session is to build a solid mental model of Spark so that you not only write working code, but efficient code and avoid burning money.

We showcase this with Microsoft Fabric Notebooks but the knowledge is applicable to any spark cluster.

• Danny Kruge

Is This Just Real-Time… or Is It Just Fabric?

Everyone wants real-time data, but what does that actually mean? When your business users say real-time, are they talking about a few seconds, a few minutes, or just “faster than yesterday”?

In this session, we’ll pull back the curtain on the buzzword and explore what “real-time” really means from near-real-time micro-batching to true event-driven streaming and how you can build these patterns directly within Microsoft Fabric.

You’ll see practical examples of how to:

Ingest and process streaming data in real time using Fabric

Compare performance between small batch pipelines and traditional data warehouse ingestion

Choose the right architecture for your organization’s definition of “real-time”

Through live demos and relatable use cases, you’ll walk away knowing how to design a Real-Time Intelligence Fabric that fits your needs and how to stop chasing fantasy real-time.

Slide deck etc

Real Time Fabric - Danny Kruge.pdf

• Gianluca Sartori

Questioning My SQL Server Faith… So You Don’t Have To

SQL Server is facing fierce competition from open-source databases like PostgreSQL, and even long-time advocates, including former MVPs and SQL Server celebrities, are recommending alternatives. Even I, under this mounting pressure, find myself questioning my long-established SQL Server faith. So if you’re starting a new project, is SQL Server still the right choice?

In this session, we’ll explore the features that set SQL Server apart, from performance and scalability to its top-notch tooling. We’ll compare these strengths against what open-source platforms have to offer, so you’ll walk away with the insights you need to confidently choose SQL Server for your next application.

• Uwe Ricken

The strengths of partitioning – practical examples

Partitioning is an advanced feature in Microsoft SQL Server that offers significant performance and manageability benefits. This session is tailored for experienced SQL Server DBAs looking to deepen their understanding of partitioning by exploring real-world showcases that highlight its transformative impact.

We will delve into two to four detailed case studies where partitioning has been successfully implemented to solve complex database challenges:

1. Accelerating a High Transactional OLTP production environment by up to 30% in a manufacturing company by partitioning machine data.

2. Optimizing the load of more than 10.000 cash register data into a big Data Warehouse of a retail company

3. recalculation of financial risks of a bank for daily reports for supervisory authorities

4. A financial services company leveraged partitioning to streamline their data archiving process, enabling faster data retrieval and compliance with regulatory requirements.

• Peter van den Bos

Practical Data Modeling in Power BI: Beyond the Star Schema

This session focuses on data modeling with a strong emphasis on real-world scenarios. A solid foundation is assumed: familiarity with the star schema, the importance of correct data types, and the core principles of a well-designed Power BI data model.

During the session, I go beyond theory and show how to make practical design choices when building fact tables. For example, when should you use a snapshot fact table? What types of business questions does it support, and how do you design such a structure correctly?

Slowly Changing Dimensions Type 2 are also covered. I demonstrate how to set up an SCD Type 2 dimension and, more importantly, how to create correct and reliable DAX calculations on top of it. The focus is not only on the technical implementation, but also on the reasoning behind these choices.

Many-to-many relationships are another key topic. What happens when data exists at different levels of granularity? When do issues arise? And is a bridge table always the right solution, or are there alternative approaches, for example when dealing with budget data?

In short, this session goes beyond the classic star schema. It presents practical solutions for realistic modeling challenges, because customer data models are rarely as clean and straightforward as the well-known Contoso example.

• Štěpán Rešl

Why Your Data Team Needs an AI Agent (Not a Chatbot)

"Just ask ChatGPT" has become the default answer for a lot of people for everything. But a one-shot prompt followed by copy-paste is not how you scale a data team's work. In this session, I'll introduce the concept of AI agents - not as a futuristic vision, but as a tool that works today within the Microsoft Fabric ecosystem.

I'll demonstrate how Claude, acting as an agent, plans, uses tools, and iterates on tasks: generating and debugging PySpark/Python notebooks, analyzing and creating Power BI reports using Rui Romano's plugins, building pipeline logic, and performing QA checks on code. All shown on live demos from real-world projects.

You'll walk away with a clear understanding of where AI agents save the most time, where they still fall short, and how to get started on your team - without big investments and without illusions.

Slide deck etc

Why Your Data Team Needs an AI Agent (Not a Chatbot).pdf

• Mark Streutker
• Niels Nagle

The Future of Data Engineering and Data Platforms in the Age of AI and Agentic Systems

AI is rapidly automating much of the traditional work performed by data engineers.
Pipelines are built faster, code is generated automatically, and operational tasks are increasingly moving into the background. Does this mean the need for data engineering is declining?

A small spoiler: we believe the exact opposite is true.

As AI automates execution, the demand for reliable data foundations, semantic models, governance, and business context continues to grow. Organizations are discovering that the biggest obstacle to successful AI adoption is not the models themselves, but the quality, accessibility, and trustworthiness of their data.
In this session, we explore how data engineering is evolving from building pipelines to designing the enterprise memory: a platform where data is made understandable, trustworthy, and usable for both humans and AI systems. We will touch upon the concept of AI-ready data and provide practical guidance on how organizations can start their journey with a single use case and scale from there. With example of an use case for conversational analytics on your semantic model.
AI is changing how we build data platforms. This session explores what that means for the future role of data engineering.

• Reitse Eskens

SQL 2025, SSMS 22 and Fabric. And they mirrored happily ever after?

You know you have your data in your database. You know how to query it. But you can't let everyone in on your production database and run their queries. Yet everyone wants to create reports or dig into the data to gain insights.

SQL Server 2025 and Fabric offer the option to mirror your database data directly into Microsoft Fabric. This offloads the analytics from your on-premises environment to the cloud.

Let me show you how this works, what caveats there are and how to set up security.
Because you don't want anyone snooping around, now do you?

After the session, you'll have a better idea of how this integration works and if this is a good fit for your situation.

Slide deck etc

SQL SSMS Fabric Mirrored happily ever after.pdf

• Christian Peeters

T-SQL Meets AI: Unleashing SQL Server 2025’s Vector & RAG Power

SQL Server 2025 introduces powerful AI capabilities that bring intelligence directly into the database engine. With native vector support and seamless connections to external AI models, developers can now build Retrieval-Augmented Generation (RAG) pipelines and natural language interfaces using only T-SQL, while keeping sensitive data under your control.

In this session, you’ll learn how to:
- Chunk and enrich your data for RAG workflows
- Generate and store embeddings in SQL Server with the new VECTOR data type
- Perform semantic vector searches to power smarter applications
- Call external AI models directly from T-SQL
- Ensure trust by keeping data securely within your systems, minimizing exposure to external services

By the end, you’ll have the practical knowledge to combine relational data and AI to build next-generation applications. All with the assurance that your data remains under your control.

Slide deck etc

TSQLmeetsAI - share.pdf

• Torsten Strauß

High-Concurrency Insert Patterns: Eliminating Page Latch Contention in SQL Server

High-concurrency workloads often expose a critical scalability challenge in SQL Server: page latch contention caused by inserts into clustered indexes with sequential key values. When many sessions attempt to insert into the same “last page” of an index, SQL Server must serialize access using latches—quickly turning into a major bottleneck in high-throughput systems.
In this session, we explore why this contention occurs and how it impacts scalability in modern workloads. Using practical demonstrations and benchmarks, we analyze four proven techniques to mitigate and eliminate page latch contention. Rather than promoting a single “silver bullet”, the session compares multiple approaches and explains the trade-offs of each design.
You will learn:
• Why sequential inserts can cause PAGELATCH contention on index leaf pages
• How to identify page latch contention using DMVs and wait statistics
• How techniques such as hash-partitioned tables and reserved key ranges distribute insert load
• When memory-optimized tables (In-Memory OLTP) can eliminate latch contention entirely
• How the OPTIMIZE_FOR_SEQUENTIAL_KEY index option improves scalability in modern SQL Server versions
Using comparative benchmarks and real-world scenarios, the session focuses on helping you choose the right strategy for your workload rather than applying a one-size-fits-all solution.
By the end of the session, you will have a clear understanding of the available techniques - and the confidence to design clustered indexes that support scalable, high-throughput insert workloads.

• Thomas Kejser

Understanding and Comparing Query Optimisers

Ever wondered if your skills in one database transfer easily to another?

In this session, I will give you a conceptual framework for thinking about SQL databases. We will look at how parsers and query optimisers work (in all databases) and how they interact with execution engines.

We will also look at SQL Arena - a tool for learning about query optimisation which also allows you to compare databases with each other.

This is a 1 slide every 2 minute session. Grab your coffee and strap in for fast ride.

• Mehmet Bozan
• Alper Teke
• Rocky Kalloe

Sovereign Data & AI: Control, Choice and Innovation in the Age of AI

Most organisations want to do more with AI, yet their most valuable data stays off-limits because they can't fully control where it goes.
Sovereignty is usually seen as a brake on innovation. We think it's the accelerator.
In this session we unpack what data sovereignty really means (hint: it's more than hosting in Europe).

• Tonie Huizer
• Peter Kruis

Two Developers, One Mission: Make a Test Database That Doesn’t Suck

If you've ever been frustrated by a test database filled with nonsense, duplicates, sensitive information, or YOLO-style restored production backups, this session is for you.
Join Peter Kruis and Tonie Huizer, two energetic data enthusiasts, on an epic (and slightly chaotic) quest to build test databases that are safe, realistic, and actually useful for development and testing. With decades of combined experience, they’ll take you on a journey through the wild world of test data, showing just how bad it can get, and how good it can be.
Through engaging role play, real-world examples, and live demos, they’ll expose the dangers of shady test environments and share smarter, safer ways to manage test data. Audience participation is encouraged; because (bad) test databases are a team sport!
You’ll leave entertained, but more importantly, you’ll leave with tools and strategies to stop your test database from being the weakest link in your development chain.
Because let’s face it: your test database shouldn’t suck.

Slide deck etc

demo-on-github.txt

• Marnix Jansen

Write Reusable DAX with User Defined Functions

In many Power BI models, the same DAX logic is repeated across multiple measures. Over time this leads to duplication, inconsistencies, and models that become increasingly difficult to maintain.

User Defined Functions (UDFs) introduce a powerful new capability to DAX: the ability to encapsulate logic once and reuse it across your model. Instead of rewriting complex calculations again and again, you can define a function once and call it wherever you need it.

In this session we will explore how UDFs work in practice and how they can significantly improve the structure and maintainability of your semantic models. We will start with the fundamentals of UDF syntax and the role of parameter types, subtypes, and passing modes.

From there we will walk through practical examples such as reusable time intelligence calculations, formatting logic, and record-counting patterns. Along the way we will also examine how evaluation context and filter context influence the behavior of UDFs.

Finally, we will discuss when UDFs are the right tool and when alternatives such as calculation groups might be more appropriate. We will also look at how emerging community libraries like DAX Lib make it possible to create reusable function packages that can be shared across models.

After this session you will understand how to design reusable DAX logic and apply User Defined Functions to build cleaner, more maintainable Power BI models.

Key Takeaways

*Understand the syntax and mechanics of User Defined Functions in DAX

*Learn how types, subtypes, and passing modes influence evaluation

*See practical patterns for reusable calculations

*Understand the difference between UDFs and calculation groups

*Discover how to build reusable DAX libraries across models

• Thijs Baan

Cognitive Shift: How Agentic AI Changes the Way We Think

Copilots and agentic AI are not only changing how we work as software engineers, but also how we think. In this session, we look beyond productivity claims and tool comparisons, and focus on the psychological side of AI Augmented Engineering. What happens to your attention, working memory, critical thinking, and mental models when AI starts to take over more of the writing, searching, and thinking?

In this talk, we explore why working with AI often feels faster and smoother, while at the same time creating new kinds of mental friction. As AI increasingly provides output, suggestions, and next steps, the role of the engineer shifts: from building things yourself to evaluating, choosing, checking, and steering. This directly affects focus, understanding, learning, debugging, and keeping track of context. Using recognizable real-world situations, we look at where AI creates mental space, where it adds cognitive load, and what this means for how engineers develop expertise.

This session gives you a practical psychological framework to use AI more consciously and effectively in your daily work. You’ll learn how to recognize when AI improves speed, focus, and mental space, when it quietly adds extra cognitive load, and how to deal with that in a smart way. This helps you get more out of copilots and agents, without losing sharpness, understanding, and control, as AI becomes an increasingly essential part of engineering.

Slide deck etc

20261010 - Data Saturday Holland.pdf

• Simon Whiteley

Giving the Business Their Vibes Back

Your users have been building apps for years: the spreadsheet that secretly runs your entire finance division, the cryptic Power App POC that quietly became production. Every attempt to stop them has failed, because the need was always real and the alternative was static dashboards. Databricks Apps changes the terms of that argument, and Genie App Builder keeps it in their hands. Building an app has gone from a project to an afternoon, and the result now sits inside the governance boundary rather than on somebody's laptop. The question facing platform teams is no longer whether the business should build. It is what happens to the hundred apps that arrive next, and who is keeping them working in six months time.

We start with what a production app on Databricks actually needs: Lakebase for state and session handling, identity and user context, permissions inherited from Unity Catalog. We then dive into Genie App Builder, looking at the future of business-owned app dev. From there we move to the fully engineered end of the scale, using agentic coding to build a serious application, and walk through the architecture of a master data platform running in production today. We close on what follows adoption: app sprawl, the ownership problem it creates, the compute and storage bill nobody forecast, and how Genie App Spaces is shaping up to close that gap.

This session is for data engineers, architects and platform owners who can see this wave coming and would rather shape it than stand in front of it. You will leave able to tell the difference between an app that survives production and one that becomes somebody else's problem, with a clear view of where the guardrails need to sit before the sprawl starts, and a strong case for building more apps rather than fewer.