Free DP-300 practice test: 30 questions with answers

A Microsoft Entra administrator with Entra authentication enables individual, auditable sign-in with corporate identities. Shared logins lose accountability, and disabling authentication or publishing passwords are unsafe.

Auditing that sends logs to a durable destination records access and changes for review. Disabling logs, auditing only restarts, and tables users can edit do not provide trustworthy evidence, and masking hides data rather than recording activity.

Transparent data encryption encrypts data at rest with platform-managed keys and requires no application changes. Client-side encryption changes application behavior, plain text is unprotected, and hashing columns breaks their usability.

A private endpoint with public access disabled restricts connectivity to the private network. Open firewall rules, public addresses, and publicizing the server name expose the database.

Dynamic data masking hides sensitive values from users without the right privilege while allowing troubleshooting on structure. Administrative rights, printouts, and shared passwords expose the data.

The vCore model exposes compute and storage separately and supports Azure Hybrid Benefit for licensing. DTU bundles limit that control, and free tiers and file shares are not production database hosting models.

Elastic pools share capacity among databases with uneven usage, which lowers cost while covering peaks. Dedicated maximum capacity for each database and manual scaling waste resources or respond too slowly.

An online migration with continuous replication keeps the target in sync until cutover, minimizing downtime. A week-long stop, spreadsheet export, and retyping the schema are slow and error-prone.

A serverless tier scales compute within limits and pauses when idle, which matches intermittent workloads. Fixed maximum tiers pay for idle capacity, while files and spreadsheets are not transactional database platforms.

Managed Instance provides near-complete SQL Server compatibility, including SQL Agent and cross-database features. A single database lacks several of those features, and table storage and SQLite are different products.

Auto-failover groups provide automatic failover to a secondary region with listener endpoints that stay constant. Weekly copies, read-only replicas without failover, and file exports do not meet the requirement.

Point-in-time restore recovers the database to a specific moment within the retention window, which fits this need. Old backups and manual re-entry lose recent data, and an empty database contains nothing to recover.

Read replicas and geo-replicas can serve read-only queries, offloading the primary. Writes cannot be sent to a read replica, and deleting read traffic or disabling indexes does not offload anything.

Zone redundancy keeps service running through a zone failure, and automated backups enable recovery. A single rack, disabled backups, and disabled monitoring each reduce resilience or visibility.

Active geo-replication provides a readable secondary in another region without automatic failover. A failover group switches roles automatically, and exports and local backups do not provide a live readable copy.

Suggested indexes must be evaluated against the real workload because each index adds write cost. Automatic creation can produce many unnecessary indexes, and deleting data or restarting does not fix the query.

Short transactions release locks sooner, and an appropriate isolation level limits unnecessary lock holding. Transactions held during user input, table-wide locks, and disabled indexes all increase blocking or hurt performance.

Forcing the previously good plan restores the known performance while the root cause is investigated. Deleting statistics can make things worse, restarts do not fix plan choice, and more users add load.

Updating statistics refreshes the information the optimizer uses to choose plans. Deleting tables, disabling the optimizer, and adding waits do not correct stale estimates.

Automatic tuning analyzes workload behavior and recommends or applies safe fixes, with rollback when a change does not help. Manual index deletion, disabled statistics, and maximum sizing do not tune the workload.

Query Store records query runtime statistics and plan history over time, which shows heavy queries and plan changes. File size, application logs, and time zone do not capture query execution history.

A metric alert rule with threshold and time window notifies the team automatically when the condition holds. Manual checks and monthly printouts react too late, and ignoring CPU overlooks a real symptom.

Extended events capture detailed engine events with low overhead, which suits rare failure analysis. Browser tools, calculators, and wallpapers do not observe database engine behavior.

Wait statistics in the dynamic management views show where time is spent waiting, such as on locks or I/O. Backup names, login lists, and firmware versions do not explain query waits.

Diagnostic settings route logs and metrics to a central workspace where they can be queried across resources. Printing, local-only logs, and disabling diagnostics do not provide central analysis.

Elastic jobs or a scheduled automation runbook run maintenance across databases on a schedule, which avoids manual work. Manual runs depend on memory, and business-hour execution affects users.

Infrastructure as code templates applied through a pipeline produce consistent environments that can be reviewed. Manual portal steps, screenshots, and memory-based setup produce drift.

An event-driven alert with an action group notifies the team as soon as the job fails. Monthly checks, no notification, and unread files delay or prevent response.

Scripts and templates create identical resources consistently and can be rerun. Manual creation is slow and error-prone, and copying files or letting users choose settings produces inconsistency.

A scheduled automation changes the tier on time without manual steps. Manual changes depend on people, maximum sizing wastes money, and waiting for complaints delays scaling.