Home / data-analytics / Administering Microsoft Azure SQL Solutions (DP-300) / Cheat Sheet
Administering Microsoft Azure SQL Solutions (DP-300)

DP-300 Cheat Sheet

DP-300 Tests Matching Each Task to the Right Azure SQL Platform

For DBAs running SQL Server workloads on Azure SQL Database, Managed Instance and Azure VMs. Many wrong answers name a real feature that exists only on another platform, so identify the target first.

Check Your Readiness →
Among the harder certs
Avg: Not published by Microsoft
Pass: 750 / 1000
Most candidates understand Administering Microsoft Azure SQL Solutions (DP-300) concepts — and still fail. This exam tests how you apply knowledge under pressure.

DP-300 Domain Map (skills measured as of October 27, 2026)

Most items are platform-sensitive: the same goal, such as scheduling, firewalling or cross-region DR, has a different answer on Azure SQL Database, Managed Instance and SQL Server on VMs. Expect RPO and RTO figures, T-SQL and DMV output to interpret. 100 minutes; question count not published.

  1. 01
    Plan and implement data platform resources (15–20%) — deployment options, service tiers, scaling, migration methods
  2. 02
    Implement a secure environment (20–25%) — Entra authentication, TDE, Always Encrypted, firewall rules, auditing, ledger
  3. 03
    Monitor, configure and optimize database resources (20–25%) — Query Store, DMVs, Extended Events, blocking, indexes
  4. 04
    Configure and manage automation of tasks (15–20%) — SQL Server Agent, elastic jobs, Bicep and ARM, alerts
  5. 05
    Plan and configure HA/DR (20–25%) — backups and PITR, LTR, geo-replication, failover groups, availability groups

Wrong instinct vs correct approach

A 4 TB on-premises database must move to Managed Instance with only minutes of downtime
✕ Wrong instinct

Export a BACPAC and import it into Managed Instance

✓ Correct approach

BACPAC is offline and slow at terabyte scale. Use the Managed Instance link, built on distributed availability groups, or Log Replay Service restoring a backup chain from blob storage, then cut over with the final log.

Equality lookups must still work on a column encrypted so DBAs cannot read it
✕ Wrong instinct

Use randomized Always Encrypted on the column

✓ Correct approach

Randomized encryption blocks equality comparisons. Choose deterministic Always Encrypted for equality filters and joins, or enable secure enclaves for range queries and pattern matching. Enclaves still never expose the keys to DBAs.

Sessions pile up waiting, and the DMV output shows blocking_session_id values
✕ Wrong instinct

Kill the session with the longest wait time

✓ Correct approach

Follow blocking_session_id to the head blocker, the session holding the locks, and resolve it; ending waiters leaves the chain intact. Read committed snapshot, on by default in Azure SQL Database, already stops readers blocking writers.

An app on SQL Server VMs must reconnect automatically after an availability group failover
✕ Wrong instinct

Point the connection string at the primary server name

✓ Correct approach

Create an availability group listener so clients follow the new primary. On Azure VMs, plan cluster, quorum and listener connectivity together; a failover cluster instance shares one storage copy and offers no readable secondary.

A Managed Instance must accept connections only from approved networks
✕ Wrong instinct

Add a server-level IP firewall rule to the instance

✓ Correct approach

Managed Instance has no server-level IP firewall rules. Access is controlled by its VNet subnet, NSGs and the choice of private or public endpoint. Server and database firewall rules belong to Azure SQL Database logical servers.

Know these cold

  • ▸ SQL Server Agent on VMs and Managed Instance; elastic jobs for Azure SQL Database
  • ▸ PITR keeps 1–35 days and restores to a new database; LTR keeps backups up to 10 years
  • ▸ Active geo-replication is Azure SQL Database only; failover groups cover Database and Managed Instance
  • ▸ User backups on Managed Instance must be COPY_ONLY to protect the backup chain
  • ▸ Deterministic encryption allows equality; randomized needs secure enclaves for any comparison
  • ▸ sys.dm_db_resource_stats holds about an hour at 15-second grain; use sys.resource_stats for history
  • ▸ Resource Governor works on VMs and Managed Instance only, never in Azure SQL Database
  • ▸ Automatic index tuning is Azure SQL Database only; Managed Instance gets plan correction
  • ▸ Serverless with auto-pause for intermittent load; elastic pools when database peaks do not overlap
  • ▸ SQL VM log disks use no host caching; ARM complete mode deletes unlisted resources

Can you answer these without checking your notes?

In this scenario: "A 4 TB on-premises database must move to Managed Instance with only minutes of downtime" — what should you do first?
BACPAC is offline and slow at terabyte scale. Use the Managed Instance link, built on distributed availability groups, or Log Replay Service restoring a backup chain from blob storage, then cut over with the final log.
In this scenario: "Equality lookups must still work on a column encrypted so DBAs cannot read it" — what should you do first?
Randomized encryption blocks equality comparisons. Choose deterministic Always Encrypted for equality filters and joins, or enable secure enclaves for range queries and pattern matching. Enclaves still never expose the keys to DBAs.
In this scenario: "Sessions pile up waiting, and the DMV output shows blocking_session_id values" — what should you do first?
Follow blocking_session_id to the head blocker, the session holding the locks, and resolve it; ending waiters leaves the chain intact. Read committed snapshot, on by default in Azure SQL Database, already stops readers blocking writers.

Common Exam Mistakes — What candidates get wrong

Proposing SQL Server Agent for jobs on Azure SQL Database

Agent exists only on Managed Instance and SQL Server on VMs. Azure SQL Database uses an elastic job agent with target groups, whose identity needs a user and permissions in every target. On Managed Instance, built-in Agent beats elastic jobs.

Choosing active geo-replication for Azure SQL Managed Instance

Active geo-replication, with up to four readable secondaries, is an Azure SQL Database feature. Managed Instance uses failover groups for cross-region DR, which also provide stable listener endpoints; logins and Agent jobs on the secondary must be created separately.

Using TDE or masking to hide data from administrators

TDE encrypts data, log and backup files at rest, but queries still return plaintext. Dynamic data masking is bypassed by administrators and UNMASK holders. Keeping values from DBAs and cloud operators needs Always Encrypted, where keys never reach the engine.

Adding vCores when CPU spikes right after a deployment

A sudden rise after a release or statistics update usually means a plan regression. Query Store shows the earlier good plan and can force it; automatic tuning with FORCE_LAST_GOOD_PLAN, on by default, applies that correction for you.

Raising point-in-time retention to meet a seven-year compliance rule

PITR retention runs 1 to 35 days, and every restore creates a new database rather than overwriting. Multi-year requirements use long-term retention, which keeps weekly, monthly or yearly full backups for up to 10 years per database.

Resources to help you pass Azure SQL Administrator DP-300 on your first attempt

See whether you pick the right Azure SQL platform feature every time on DP-300.