Dataset backup and restore performance just got a lot better

In August 2021, we announced general availability (GA) of backup and restore for large datasets. Not content, we looked for additional ways to improve this important feature. One area stood out: performance. After all, the faster you can backup and restore your datasets in Power BI, the more you can minimize downtime in critical situations and uphold your service level agreements (SLAs). Of course, better performance also helps with backup and restore based migrations of Azure Analysis Services (AAS) datasets to Power BI.

Today, we’re happy to announce significant performance improvements to backup and restore operations. Our benchmark tests with 1GB, 5GB, and 10GB datasets show backups can now be completed up to 22 times faster than before, and the time required for restore operations can also be reduced by more than 30%.

With the goal of improving the speed of backup and restore for large datasets, we evaluated several critical areas in the Analysis Services engine. The Analysis Services engine is the key component in Power BI that loads and processes datasets and answers analytical queries. It’s also the key component in backup and restore operations. Our evaluation showed that datasets over 1GB offered significant improvement potential. This potential was even greater with increasing dataset sizes. You can see this in the chart below showing the backup duration in seconds (DurationSec) by dataset size (DB Size).

The backup process in-particular exhibited a non-linear duration as the dataset size grew. At a dataset size of 30GB, we finally reached a threshold that compelled us to prioritize making changes to the backup logic.

Restore operations performance also improved. As shown below, our evaluations showed a 33% performance improvement, which is a substantial gain when you need to urgently recover a database.

These backup and restore performance improvements are now available in Power BI without requiring any configuration changes or optimizations to your datasets. If you notice that your backups complete in a fraction of the time it took previously, rest assured your backups did complete successfully. They just completed faster.

And that’s it for this quick announcement of these exciting backup and restore performance improvements in Power BI. Be sure to check out Backup and restore datasets with Power BI Premium to learn more about how to use dataset backup and restore and multi-geo considerations.

Comments

Post a Comment

Popular posts from this blog

Failed to execute the package or element. Build errors were encountered

Restore of database 'DataBase_Name' failed. (Microsoft.SqlServer.Management.RelationalEngineTasks)

On-premises data gateway December release