1. Fast Analysis with Direct Lake
One of the benefits of using Fabric to create semantic models is access to Direct Lake, a fast way to load data directly into PowerBI. In this video, we’ll compare Direct Lake to the traditional ways of loading your data.
2. Previous methods of loading data (Part 1)
Before working with Fabric, there were two main ways of loading data into PowerBI.
The first was through Direct Query mode. The PowerBI engine queried the data directly from the source when using Direct Query mode. This was fairly slow; DAX queries written in PowerBI had to be translated to other query languages that could work with the original source data.
That being said, Direct Query mode had the benefit of always being up to date. Because it queried the data directly, any changes in the data would be immediately reflected in PowerBI.
3. Previous methods of loading data (Part 2)
The other way of accessing data in PowerBI was through import mode. When using import mode, PowerBI created a cached copy of the source data. This resulted in relatively fast queries, but the data was not necessarily current. If any updates were made to the original source data, those updates would not be reflected in your PowerBI analysis unless you intentionally refreshed your model.
It's worth noting that with import mode you could schedule refreshes; you weren't dependent on manually refreshing the data. But regardless, you could be working with out-of-date data depending on when your refresh happened.
In summary, there are drawbacks to both methods
4. Direct Lake mode
This is where Fabric comes in. When you create a semantic model via Fabric, PowerBI connects to the original source through a method called Direct Lake.
Direct Lake mode is the best of both worlds; it is fast, and it uses up-to-date data. This is accomplished by using the parquet files found in OneLake. This is yet another example of the benefits of Fabric’s fundamental goal of providing a single unified data source. By using OneLake, Fabric is able to improve the PowerBI experience.
Take a look at this diagram from Microsoft comparing the three different modes. On the right hand side of the image, you see an analyst preparing to use Power BI.
5. Direct Lake mode
They have the option of choosing the slow, but real-time DirectQuery mode
6. Direct Lake mode
the fast but latent Import mode,
7. Direct Lake mode
and the best-of-both-worlds option of Direct Lake mode.
If you’re a PowerBI user who frequently works with very large datasets, or datasets that are frequently updated, try creating a semantic model in Fabric to see the benefits of working with Direct Lake.
8. Let's practice!
Let’s jump back into Fabric to see how Direct Lake works!