Parameters Power BI Import Power BI DirectQuery
Performance It has a high-performance query engine, and the data model is already cached with no latency, but with more volume, it may degrade. Since queries are processed in real-time, network connectivity is crucial. If indexes are well created, they perform well.
DAX Expressions and Transformations Supports all DAX functions and advanced Power Query transformations for shaping the data before importing it Restricted support for DAX functions and transformations since you can use limited data available in the data source
Schedule Refresh Hourly/daily programmed async jobs with a maximum of eight schedules per day Schedule at every 15-minute interval or in real-time
Size Approximately 1GB per dataset/model No limit to storing data in an on-premises database
Source of Data Importing data from multiple sources Importing data from a single source
Data Modeling No limitations Some limitations
In-built Hierarchy Accessible Not accessible
Data Storage Data is preserved in the Power BI Service since it is in the cloud The cloud service will not store any data; it will be stored locally
Data Availability If data refresh doesn’t work, the last data available in the model is used If data refresh doesn’t work, the report goes black since data is not stored
Change in Data Connectivity Mode Cannot change from Import to Direct Query The mode can be changed from Direct Query to Import
Target Groups Small to Medium Datasets Large Datasets (greater than 1 GB)
Security Users can create row-level security on the PBI dataset Users can reuse on-premises row-level security for Analysis Services
Clustering, Quick Insights, Calculated Tables Available Not available
Flexibility and Control There is more control over data; hence, creating complex data models is more accessible but may not be suitable for larger datasets. Since data is not stored in Power BI, it is not that flexible in creating complex data models, but it can easily manage large datasets.