DP-600 Cheat Sheet: Key Facts to Review

The short version

Use this sheet in the last days before the exam, after you have studied the material. It lists the exam facts, the ideas that decide many questions, and the pairs of terms people mix up. It is a revision aid in our own words. It does not cover everything on the exam, and Microsoft’s study guide stays the authority.

Exam facts at a glance

Item	Fact (checked October 22, 2026)
Exam	DP-600, Implementing Analytics Solutions Using Microsoft Fabric
Time	100 minutes
Pass mark	700 on a scale of 1 to 1,000
Questions	Not published
Domains	Maintain (25-30%), Prepare data (45-50%), Semantic models (25-30%)
Outline	Skills measured as of October 19, 2026
Tools	You may open Microsoft Learn; the clock keeps running
Languages	Seven, English updated first

Domain 1: Maintain a data analytics solution

Key concepts

Access works in layers: workspace role, item permission, then rules inside the data (row, column, object, file).
Sensitivity labels classify and protect content. Endorsement signals which items to trust.
Deployment pipelines move content between stages. Deployment rules change settings per stage.
Git integration versions a workspace. A .pbip project stores a Power BI project as files.
The XMLA endpoint lets tools deploy and manage semantic models.

Confused pairs

Workspace role is for broad access. Item permission is for sharing one thing.
Deployment rule is for per-stage settings. A Git branch is for parallel development.
Sensitivity label is about protection. Endorsement is about trust.
.pbit is a template. .pbids points to a data source.

Decision rules

If one person needs one report, think item permission, not a workspace role.
If the stem says “different connection in production”, think deployment rule.
If the stem asks what breaks downstream, think impact analysis.

Domain 2: Prepare data

Key concepts

Stores: lakehouse for files and Spark, warehouse for T-SQL workloads, eventhouse for streaming and event data.
A shortcut references data in place. Copying creates a second set to maintain.
Dataflow Gen2 suits low-code transformation, notebooks suit code and scale, pipelines orchestrate and copy, T-SQL fits warehouse logic.
Star schema: facts hold measures and keys, dimensions describe them.
Clean data means handling duplicates, nulls, and wrong types before modeling.

Confused pairs

Lakehouse versus warehouse: Spark and open files versus T-SQL engine.
Shortcut versus mirroring: reference versus a continuously replicated copy.
Star versus snowflake: one level of dimensions versus normalized dimensions.
SQL versus KQL: tables and joins versus pipes read left to right.

Decision rules

Team knows T-SQL and needs multi-table transactions: warehouse.
Mixed file types and Spark: lakehouse.
Event streams and time-based queries: eventhouse with KQL.
Need the latest value per key in KQL: think `arg_max`.

Domain 3: Implement and manage semantic models

Key concepts

Storage modes: Import is fast but needs refresh, DirectQuery asks the source each time, Direct Lake reads Delta tables in OneLake.
Relationships: one-to-many is the default. Many-to-many and bridge tables solve special cases.
DAX: variables improve clarity, iterators work row by row, `CALCULATE` changes filter context.
Calculation groups, field parameters, and dynamic format strings reduce repeated measures.
Incremental refresh loads only recent partitions.

Confused pairs

Filter context versus row context.
Direct Lake on OneLake versus Direct Lake on the SQL analytics endpoint.
Composite model versus a single storage mode.
Calculated column (stored, computed at refresh) versus measure (computed at query time).

Decision rules

Huge data and freshness without copying: think Direct Lake.
Slow visual: check the model and DAX before adding hardware.
One measure needed for several time calculations: think calculation group.

Common traps

Picking the most powerful tool instead of the simplest that meets the stated need.
Ignoring a single constraint in the stem, such as “read-only”, “no code”, or “least privilege”.
Treating Direct Lake as if it behaved exactly like Import.
Bi-directional relationships used by default.
Missing that the question asks for two answers.

Last-day checklist

Re-read the weak points from your latest practice results.
Skim the confused pairs above out loud.
Open the exam sandbox once, so the screens feel familiar.
Check your ID, booking time, and system test for online delivery.
Sleep. Do not start new topics.

More guides for DP-600

Next step

Check your recall with the sample questions, then read the exam format page for the rules of the day. For a full run under time pressure, use the DP-600 practice exams. They are a readiness check and do not replace hands-on time in Fabric.

Try the DP-600 practice exams