Matching a workload to the right Fabric data store
Microsoft Fabric provides multiple data store options (Lakehouse, Warehouse, and specialized stores like KQL Database and Eventhouse), each optimized for different data types, workloads, and skill sets. Choosing the right store depends on whether you need file-based flexibility, SQL-based structured analytics, or real-time analytical queries over streaming/log data.
1 · Learn the must-know
- A Lakehouse stores data as files (Delta/Parquet) in OneLake and supports both structured tables and unstructured files, making it ideal for data engineering, data science, and Spark-based workloads.
- A Warehouse provides a fully relational, T-SQL-based experience with full DDL/DML support and is best suited for traditional data warehousing, BI reporting, and teams with strong SQL skills.
- Both Lakehouse and Warehouse store data in Delta Lake format in OneLake, enabling cross-engine querying (e.g., a Warehouse's SQL endpoint can query Lakehouse tables and vice versa) without data duplication.
- KQL Database (within an Eventhouse) is optimized for high-velocity, time-series, and semi-structured data such as logs, telemetry, and IoT streams, using Kusto Query Language (KQL) for near real-time analytics.
- Lakehouse SQL endpoints are read-only for querying Delta tables via T-SQL, while write operations to Lakehouse tables must go through Spark, pipelines, or dataflows, not the SQL endpoint.
- Choice of data store should be driven by workload pattern (batch vs. streaming), query language preference (Spark/PySpark vs. T-SQL vs. KQL), governance/security needs, and downstream consumption (Power BI, notebooks, or client apps).
2 · Check your understanding
A data engineering team at a retail company must maintain a curated dimensional model that is refreshed nightly using ELT logic requiring multi-statement T-SQL transactions, including INSERT, UPDATE, and DELETE operations across several related tables. The resulting model is consumed directly by Power BI reports through T-SQL queries. Which Fabric data store should the team choose to host this model?
What you have tried across DP-700's objectives, not a readiness score.
Implement and manage an analytics solution30-35% of the exam0 of 18 tried
Ingest and transform data30-35% of the exam0 of 19 tried
Monitor and optimize an analytics solution30-35% of the exam0 of 17 tried
3 · Keep going
Ready for more? Take a weighted mock or try free practice questions.