Optimizing models

You have heard that a star schemas are optimal for Power BI. You have seen the dramatic effect of compression on your data models. And the storage engine makes parallelism happen? Or something like that!

Have you ever wondered *why* these things are true and how they work? Or how you can use this knowledge to make models that perform better?

This session consists of a deep dive into the drivers of performance in analytical databases, and the specific optimizations of the Vertipaq storage engine, compression, and query execution. Then, we will build upon this understanding to optimize models to work well with VertiPaq. You will learn *why* certain patterns of models are optimizations, not just what those patterns are.

This includes approximately 1/3 introduction and theory and extended worked examples and model optimizations both for total size in RAM and query performance at report time.

Coming out of this session, you will understand VertiPaq at a deeper level and be able to identify memory and performance optimizations for your models.

Attendees should be comfortable with the basics of DAX and have built multiple datasets in Power BI or Analysis Services.

Power BI operations: Monitoring and administration at scale

Building a report and deploying it is just the beginning.

This session provides an overview of operational data that you should be monitoring as a Power BI administrator, data lead, or anyone who is responsible for more than one report. We will also cover the tools available out of the box and that can be built to operate Power BI at enterprise scale.

Topics covered include:
- Refresh monitoring
- Refresh performance
- User activities and report views
- Capacity metrics and monitoring
- Data and security lineage
- Auditing user permissions

Beyond the GUI: Power Query and M techniques

A fast-paced and demo-heavy tour of advanced functionality in Power Query with case studies based on problems seen in the field with consulting clients. This session will introduce the M language through a series of demo exercises.

Coming out of this presentation, you will be familiar with M's syntax, and have fully worked examples of developing a new ETL workflow entirely in M, without the GUI.

Attendees are expected to be comfortable with Power Query, but need not be advanced.

Beyond the star: Data modeling techniques

If you ask most people what you should do for a good analytical data model, they will start and end the conversation with "Star schema". But there is a lot more. Models are tools for thought, and this presentation will help you think better about your data.

With examples of real problems and (attempted) solutions, this presentation will examine dimensional modeling techniques which simplify and speed up measure evaluation. A fast-paced look at some common and uncommon reporting scenarios, including:

- All manner of date calculations and comparisons
- Slowly changing dimensions
- Events in progress
- Conditional evaluation of measures

By the end of this talk, you should have a stronger understanding of the right questions to ask that will make your BI solution simpler for end users and more performant at run-time.

Attendees are expected to have built solutions in Power BI before, and have at least a cursory understanding of dimensional models.