Azure DP-300 flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

DTU vs vCore model

DTU bundles compute, memory, IO and storage into fixed tiers (Basic/Standard/Premium). vCore lets you choose compute and storage independently across General Purpose, Business Critical and Hyperscale, and qualifies for Azure Hybrid Benefit.

Serverless auto-pause behavior

In the General Purpose serverless tier, compute scales automatically between a minimum and maximum vCore count, and the database pauses after a configurable inactivity period of 1 hour to 7 days, billing storage only.

Elastic pool decision rule

An elastic pool shares one eDTU or vCore budget among many databases. It is cost-effective when databases have spiky peaks that do not overlap; a single database with sustained peak usage is cheaper on its own tier.

Managed Instance subnet requirement

Azure SQL Managed Instance must be deployed into a dedicated, delegated subnet with a network security group and route table. The subnet minimum is /27, and /26 is recommended to allow scaling operations.

Hyperscale tier characteristic

The Hyperscale tier separates compute from a page-server based storage layer, supports up to 100 TB per database, provides rapid restore from snapshots regardless of database size, and allows named read replicas.

Azure Hybrid Benefit usage

Azure Hybrid Benefit applies existing SQL Server licensed cores with Software Assurance to the vCore rate of Azure SQL Database, Managed Instance, or SQL on VM, reducing compute cost. It does not apply to the DTU model or serverless compute.

Contained database user

A contained user is defined inside the database with no login at the server level. Azure SQL Database uses contained users mapped to Azure AD identities or to a SQL password for authentication and permission assignment.

Private endpoint for Azure SQL

A private endpoint assigns a private IP from your virtual network to the Azure SQL server, so traffic stays on the Microsoft backbone. Setting public network access to Disabled ensures connections arrive only through private endpoints.

TDE customer-managed key prerequisites

To use a customer-managed key for Transparent Data Encryption, the Azure Key Vault must have soft delete and purge protection enabled, and the logical server must be granted key permissions via managed identity.

Deterministic vs randomized encryption

In Always Encrypted, deterministic encryption always produces the same ciphertext, enabling equality comparisons, joins and indexing. Randomized encryption is more secure but prevents comparisons and indexing on that column.

Dynamic data masking limitation

Dynamic data masking rewrites values in query results at the service layer; stored data remains plaintext. Users with the UNMASK permission, and administrators, see original values, so masking is not an encryption feature.

Row-level security implementation

Row-level security uses an inline table-valued function as a filter predicate, bound to tables by a security policy. Queries automatically return only rows satisfying the predicate for each connecting user.

Defender for SQL scope

Microsoft Defender for SQL provides threat detection alerts, anomaly detection and vulnerability assessment. It is enabled per logical server or Managed Instance and covers the databases hosted there, with per-database overrides available.

Query Store function

Query Store persists query text, execution plans and runtime statistics in the user database, enabling plan regression detection, forced plans and wait statistics. Automatic Tuning depends on it being enabled and running.

Automatic tuning FORCE PLAN

FORCE PLAN makes Query Store force the last known good plan when a regression is detected. It is enabled per server or per database with Inherit, ON, or OFF and is available only on Azure SQL Database and Managed Instance.

sys.dm_db_resource_stats usage

This view returns CPU, data IO, log IO and memory usage as a percentage of the service tier limit, sampled about every 15 seconds for roughly the last hour, making it the primary tool for tier saturation checks.

Missing index DMV caveat

sys.dm_db_missing_index_details estimates index benefit from currently cached plans only. It ignores columns not used in predicates, resets on restart, and should inform rather than dictate index design.

RESOURCE_SEMAPHORE wait meaning

RESOURCE_SEMAPHORE indicates queries are queued waiting for memory grants before execution. Common causes are overly large sorts or hash joins, missing indexes, and parameter sniffing producing oversized grants.

Database scoped MAXDOP

ALTER DATABASE SCOPED CONFIGURATION SET MAXDOP = n configures the degree of parallelism per database in Azure SQL Database, letting you tune parallelism without affecting other databases on the same server.

Resumable online index rebuild

A resumable rebuild preserves completed work when paused or interrupted, reducing log growth and downtime. It requires SQL Server 2017 or later, Azure SQL Database Business Critical, or Managed Instance.

Elastic Database Jobs

Elastic Database Jobs schedule T-SQL against Azure SQL Database targets. A job agent uses a dedicated job database, and each job runs steps against a target group containing databases, servers, or elastic pools.

SQL Agent availability

SQL Server Agent is unavailable in Azure SQL Database. It exists in Azure SQL Managed Instance and in SQL Server on Azure VM, where it can run native jobs, maintenance plans and multi-step schedules.

Elastic job target group

A target group defines which databases an elastic job runs against. Members can be an explicit list of databases, or all databases on a server or in an elastic pool, evaluated at execution time.

DACPAC deployment method

A DACPAC is a compiled schema package deployed with sqlpackage or a pipeline task. It incrementally compares the package to the target and generates the change script, and can block deployment on data loss.

Azure Automation runbook

An Azure Automation runbook runs PowerShell or Python on a schedule or webhook, often invoking T-SQL or management cmdlets. It can automate scaling, index maintenance, and administrative tasks where SQL Agent is unavailable.

Azure Monitor action group

An action group is a reusable set of notification and action targets invoked by an alert rule, such as email, SMS, voice call, webhook, Azure Function, Logic App, or Automation runbook.

Recovery point objective

RPO is the maximum acceptable amount of data loss measured in time before an outage. An RPO of 5 minutes means the DR design must replicate or back up at least every 5 minutes.

Auto-failover group listener

A failover group exposes a read-write listener endpoint and a read-only listener endpoint whose names stay constant across failover, so applications do not need connection string changes after a regional failover.

Point-in-time restore window

Azure SQL Database point-in-time restore uses automated backups and supports restoring to any point within 1 to 35 days, with 7 days as the default retention configured per database.

Long-term retention policy

Long-term retention stores full backups in Azure Blob storage on weekly, monthly, or yearly schedules for up to 10 years, independent of the short-term point-in-time restore window.

Zone redundancy SLA

Zone-redundant deployments spread replicas across three availability zones in a region, raising the availability SLA to 99.995 percent versus 99.99 percent for single-zone deployments, without changing the connection endpoint.

Geo-restore characteristics

Geo-restore creates a new database in a different region from the geo-replicated backup copy after a regional outage. It has an RPO of about one hour and no guaranteed recovery time objective.