DP-800 Study Plan: 4-Week and 8-Week

The short version

If you already write T-SQL every week, four weeks at about 10 hours a week is a realistic plan. If AI features, Git workflows, or Data API builder are new to you, take eight weeks at about 5 to 7 hours a week. These numbers are estimates, not Microsoft figures. Your own pace depends on your background, so adjust after the first week.

Before you start

Download the skills outline from the Microsoft Learn DP-800 study guide. Print it or paste it into a checklist. Every bullet is a possible question.
Set up a place to practice. A free local SQL Server developer edition, an Azure SQL database, or a SQL database in Fabric all work. Hands-on time matters more than reading on this exam.
Pick a calendar date for the exam so the plan has an end.
Decide how you will take notes. A table of “look-alike pairs” (CDC and Change Tracking, masking and encryption, ANN and KNN, full-text and vector search) is the most useful single page you can build.

Four-week plan (about 10 hours a week)

Week 1: Design and develop, part 1

Tables, data types, indexes, columnstore, constraints, sequences, partitioning.
Specialized tables: in-memory, temporal, external, ledger, graph. Create one of each in a lab and note when you would pick it.
JSON columns and indexes, and the JSON functions on the skills list.
Take practice test 1 cold at the end of the week, only to see your baseline.

Week 2: Design and develop, part 2, and security

Views, functions, stored procedures, triggers, CTEs, window functions, regular expressions, fuzzy matching, `MATCH` graph queries, error handling.
GitHub Copilot instruction files, model and MCP options, and the security impact of AI-assisted tools.
Security: Always Encrypted, column-level encryption, Dynamic Data Masking, Row-Level Security, permissions, auditing, passwordless access.

Week 3: Optimize, deploy, integrate

Isolation levels, execution plans, Query Store, DMVs, blocking and deadlocks.
SQL Database Projects, branching, pull requests, schema drift, deployment pipeline controls, secrets.
Data API builder configuration, REST and GraphQL entities, and change handling (CES, CDC, Change Tracking, Azure Functions, Logic Apps).

Week 4: AI capabilities and final review

External models, embedding design, vector types and indexes, `VECTOR_DISTANCE` and `VECTOR_SEARCH`, hybrid search, reciprocal rank fusion, and a small RAG call with `sp_invoke_external_rest_endpoint`.
Mid-week: review every domain where your practice results were weakest.
End of week: one full timed practice test, then the last-3-days list below.

Eight-week plan (about 5 to 7 hours a week)

Week	Focus
1	Tables, constraints, indexes, specialized tables; practice test 1 as a baseline
2	JSON, regular expressions, fuzzy matching, CTEs, window functions, graph queries
3	Views, functions, procedures, triggers, error handling; AI-assisted tools and Copilot files
4	Encryption, masking, RLS, permissions, auditing, endpoint security
5	Performance: plans, Query Store, isolation levels, blocking; start SQL Database Projects
6	CI/CD, pipelines, Data API builder, Azure Monitor, change handling
7	External models, embeddings, vector search, hybrid search, RRF, RAG
8	Mixed practice, weak-area review, one timed full practice test, last-3-days list

If a week goes badly, repeat it instead of squeezing the next one. A plan that finishes a week late still beats one that skips topics.

How to use practice exams in the plan

Take the first one cold, before studying much. The domain breakdown tells you where to put time.
After each attempt, read the explanation for every question, including the ones you got right. A lucky guess looks the same as knowledge.
Keep a short error log: the topic, why you missed it, and the rule that fixes it.
Use the later practice sets for different skills. Our five DP-800 practice tests start with core concepts, then move to scenarios, trade-offs, syntax details, and a long full-length set.
Retake a set only after a few days, so you are testing memory of the concept and not of the question.

Hands-on beats reading

For each domain, build something small. For example: a table with a JSON column and an index, a Row-Level Security policy, a SQL Database Project in a Git repository, a Data API builder config for two entities, and a table with an embeddings column that you query with vector search. Each takes an hour or two, and each one turns a vague bullet into something you remember.

The last three days

Three days out: take a full timed practice test. Review misses by domain, not by question.
Two days out: re-read your look-alike table and your error log. Skim the skills outline and tick what you can explain without notes.
The day before: light review only. Check your exam appointment, ID, and equipment. Sleep.
Exam day: arrive early or log in early, and read each scenario completely before the options. You have 120 minutes, so check your pace now and then and flag anything that eats time.

For the rules of exam day itself, see our exam day guide. For general study methods, see how to study for IT certification exams.

More guides for DP-800

Next step

Look at where your hours should go on the domains explained page. Then, when you are ready for a baseline, try the DP-800 practice tests.

Try the DP-800 practice exams