Luyện thi Azure Data Engineer DP-203

Blob Storage stores unstructured objects and data lake files. SQL Database is relational, Cosmos DB is NoSQL, and Databricks is analytics compute.

Event Hubs ingests millions of streaming events. SQL Database is relational, Blob is object storage, and App Service hosts web apps.

IoT Hub ingests device telemetry and supports device management. Data Lake stores files, Synapse analyzes, and Table Storage is NoSQL.

Data Factory builds pipelines that copy and transform data. DNS resolves names, CDN caches content, and Front Door routes traffic.

A data lake stores raw structured, semi-structured, and unstructured data. Relational databases, queues, and networks are different.

ELT extracts data, loads it into the target, then transforms it there. The other combinations are not the standard meaning.

Partition keys spread data across partitions to improve scale. Encryption keys, passwords, and DNS records are different.

Sinks are destinations in a copy or pipeline activity. Sources provide data, transformations change it, and triggers start runs.

Sources are where pipeline activities read data. Destinations are sinks, transformations change data, and schedules control timing.

Blob stores objects, Event Hubs ingests streams, Data Factory orchestrates, and data lakes store raw data.

Databricks provides managed Spark for data transformations. DNS, CDN, and Front Door are networking services.

Synapse SQL runs relational SQL queries and transformations. Spark is for Scala/Python, Pipelines orchestrate, and Link connects sources.

Mapping data flows provide visual transformation logic in Data Factory. Tables, dashboards, and events are different.

Notebooks let engineers run interactive code for transformation. Tables, triggers, and storage accounts are different.

Streaming queries process real-time data. Batch queries, views, and stored procedures are not streaming.

Window functions compute values over related rows. Containers, firewalls, and load balancers are unrelated.

Dimension tables store descriptive attributes such as products and customers. Fact tables store events, backups protect data, and pipelines orchestrate.

Fact tables store measurable events and references to dimensions. Dimensions describe, dashboards visualize, and notebooks run code.

Data quality measures accuracy, completeness, consistency, and reliability. Size, cost, and tier are different.

Databricks runs Spark, Synapse SQL runs SQL, mapping data flows are visual ETL, and notebooks run interactive code.

Data Factory orchestrates and schedules pipelines. Blob stores objects, DNS resolves names, and CDN caches content.

Pipelines are ordered collections of activities. Containers, VMs, and DNS zones are not pipelines.

Activities are steps such as Copy, Data Flow, or Execute Pipeline. Triggers start pipelines, and dashboards visualize.

Triggers start pipeline runs on schedules or events. Transformations, tables, and connectors are different.

Linked services define connection details such as endpoints and credentials. Tables, runs, and triggers are different.

Datasets describe data shape and location for activities. VMs, DNS records, and firewall rules are unrelated.

A run is a single execution of pipeline activities. Tiers, tables, and schedules are not runs.

Checkpoints preserve stream state so jobs can resume. Sources, accounts, and load balancers are different.

Idempotent jobs can be rerun without harmful duplication. Encryption, scaling, and scheduling are different.

Pipelines contain activities, activities perform operations, triggers start runs, and linked services define connections.

Key Vault stores secrets, keys, and certificates. Blob stores objects, SQL is relational, and CDN caches content.

Managed identities let Azure services authenticate without stored credentials. Passwords, firewall rules, and DNS records are different.

Encryption at rest protects stored data. TLS protects in transit, masking hides values, and deletion removes data.

TLS protects data in transit. At-rest encryption protects stored files, masking hides values, and backups copy data.

Private endpoints expose services inside a virtual network. Public IPs, firewall rules, and DNS records are different.

Dynamic data masking limits sensitive data exposure in query results. Deletion, encryption, and scaling are different.

Purview provides data cataloging, classification, and lineage. CDN, Front Door, and DNS are networking services.

RBAC assigns roles to control permissions. Firewalls filter traffic, backups protect data, and DNS zones resolve names.

Lineage shows data origins and transformations. Tiers, triggers, and indexes are unrelated.

Key Vault stores secrets, managed identities authenticate services, private endpoints provide private access, and Purview governs data.

Azure Monitor collects metrics, logs, and alerts for pipelines and resources. DNS, CDN, and Front Door are networking.

Metrics are time-series numeric values. Logs are events, alerts notify, and dashboards display.

Logs record events and diagnostics. Metrics are numeric, VMs are compute, and storage accounts store data.

Alerts notify teams when conditions are met. Logs, tables, and activities are not alerts.

Diagnostic settings send platform logs and metrics to storage, Log Analytics, or Event Hubs. Firewalls, backups, and DNS are different.

Cost optimization balances efficiency and performance. Deleting data, disabling monitoring, or adding storage are not optimization.

Hot and cool tiers manage access cost and retrieval. Compute sizes, pipeline modes, and roles are different.

Performance insights analyze query behavior to improve tuning. Tiers, firewalls, and backups are unrelated.

Capacity planning sizes storage and compute for demand. Deletion, key rotation, and masking are different.

Monitor collects data, alerts notify, diagnostic settings route logs, and metrics are numeric measurements.

Synapse Analytics combines SQL, Spark, and pipelines. Blob is storage, DNS resolves names, and CDN caches content.

SQL pools run relational warehouse queries. Containers store data, Spark jobs process, and pipelines orchestrate.

Spark pools run Scala, Python, and SQL analytics at scale. SQL pools are relational, storage accounts store data, and triggers start runs.

Synapse pipelines orchestrate data movement and transformation. Dashboards visualize, notebooks run code, and tables store data.

Lake databases define schemas over data lake files. SQL databases, networks, and firewalls are different.

Dedicated SQL pools have reserved compute for consistent performance. Serverless SQL is on-demand, Spark is processing, and tiers are storage.

Serverless SQL queries data lake files without provisioning. Dedicated pools are provisioned, Spark processes, and pipelines orchestrate.

Workspaces bring together SQL, Spark, pipelines, and linked data. Containers, DNS zones, and firewalls are separate.

Data flows define visual transformations in pipelines. Backups, VMs, and triggers are different.

Workspaces centralize, dedicated SQL pools are provisioned warehouses, serverless SQL queries files on demand, and Spark pools process at scale.