Azure AZ-900Trap Reference

Commonly Confused Services on AZ-900

Azure Fundamentals confusions follow the same pattern as other foundational exams: the problem is not knowing what a service does in isolation, it is missing which constraint in the question makes one service the right choice over a similar one.

Each section below gives you the deciding signal, a quick check to run when you encounter the confusion, and why the wrong answer keeps looking right.

Azure Blob StorageAzure FilesAzure Queue StorageAzure Table Storage
#1

Unstructured objects vs. shared file system vs. messaging vs. NoSQL key-value

All four are Azure Storage account services and candidates apply "Azure Storage" without distinguishing between them.

Deciding signal

Blob Storage stores unstructured data — images, documents, videos, backups — accessed over HTTP with a flat namespace. It is the right answer for storing any binary or text object. Azure Files provides fully managed SMB and NFS file shares mountable on Windows, Linux, and macOS — the right answer when the scenario involves a shared file system accessible by multiple VMs or applications using standard file protocols. Queue Storage provides simple message queuing for decoupling application components — messages up to 64 KB, retention up to 7 days. Table Storage is NoSQL key-value storage for structured, schemaless data — less capable than Cosmos DB but cheaper for simple tabular data. The signal is the access pattern: object storage (Blob), shared file system mount (Files), lightweight message queue (Queue), simple tabular NoSQL (Table).

Quick check

Is this storing unstructured files/objects (Blob), mounting a shared file system (Files), queuing messages between services (Queue), or storing simple key-value structured data cheaply (Table)?

Why it looks right

Blob Storage is the most familiar and candidates apply it to file-share scenarios where Azure Files is correct because the workload requires SMB mount or POSIX file system semantics.

Azure FunctionsAzure App ServiceAzure Container AppsAzure Kubernetes Service
#2

Event-driven serverless vs. PaaS web hosting vs. serverless containers vs. managed Kubernetes

All four run application code without requiring full VM management, so candidates treat them as interchangeable.

Deciding signal

Azure Functions is event-driven serverless compute — triggered by events (HTTP, timer, queue message) with automatic scaling and consumption-based billing. Best for short tasks. App Service is a PaaS platform for hosting web apps, REST APIs, and mobile backends in multiple languages — it manages the underlying infrastructure. Best for persistent web applications. Container Apps is a serverless containers platform — it runs containers with automatic scaling including scale-to-zero, without managing Kubernetes directly. Best for containerized microservices. AKS is managed Kubernetes — you retain full Kubernetes API control, custom operators, and cluster configuration. Best when Kubernetes expertise and full orchestration control are required.

Quick check

Is this short event-triggered functions (Azure Functions), a persistent web app without containers (App Service), containerized services with serverless scaling (Container Apps), or full Kubernetes control (AKS)?

Why it looks right

App Service is the familiar web hosting answer. Container Apps is the correct answer when the scenario involves containers with serverless scaling — and AKS is correct when Kubernetes API compatibility or operator patterns are required.

Microsoft Entra ID (Azure AD)Azure RBACSubscriptions and Management Groups
#3

Identity store vs. permission assignment vs. billing and resource boundary

All three control who can access what in Azure, so candidates blur identity, permissions, and scope.

Deciding signal

Microsoft Entra ID (formerly Azure Active Directory) is the identity provider: it stores users, groups, and service principals, and handles authentication (who you are). Azure RBAC assigns roles to identities at a specific scope (resource, resource group, subscription, management group) — it controls what authenticated identities can do. Subscriptions are billing and resource isolation boundaries; management groups are hierarchical containers for applying governance policies across subscriptions. The exam tests whether candidates know that Entra ID authenticates, RBAC authorizes, and the subscription/management group hierarchy defines the governance scope.

Quick check

Is this about storing identities and authenticating users (Entra ID), assigning permissions to control what users can do on resources (RBAC), or organizing subscriptions into a governance hierarchy (Management Groups)?

Why it looks right

RBAC and Entra ID are both "security" tools and candidates conflate authentication with authorization. Entra ID proves who you are; RBAC determines what you can do with that identity.

Azure MonitorLog AnalyticsApplication Insights
#4

Unified monitoring platform vs. log query workspace vs. application telemetry

All three appear under the Azure Monitor umbrella, so candidates treat them as the same thing.

Deciding signal

Azure Monitor is the umbrella platform that collects metrics and logs from all Azure resources and feeds them into Log Analytics workspaces or Application Insights. Log Analytics is a workspace where logs from Monitor, VMs, containers, and other sources are stored and queried using Kusto Query Language (KQL). Application Insights is an APM service specifically for application telemetry — request rates, response times, exceptions, dependencies, and user behavior. When the scenario involves querying logs from multiple Azure resources or security logs, Log Analytics. When it involves instrumenting an application to capture request traces and exceptions, Application Insights.

Quick check

Is this the platform collecting all metrics and logs (Monitor), a workspace for querying logs from any Azure resource (Log Analytics), or application-level telemetry with performance traces and exceptions (Application Insights)?

Why it looks right

Azure Monitor is the umbrella and candidates treat Log Analytics and Application Insights as equivalent options under it. They have distinct purposes: Log Analytics for platform and infrastructure logs; Application Insights for application performance and user behavior.

Azure SQL DatabaseAzure Cosmos DBAzure Table Storage
#5

Managed relational SQL vs. globally distributed multi-model NoSQL vs. cheap simple key-value

All three are database options, so candidates default to Azure SQL or Cosmos DB for any database question.

Deciding signal

Azure SQL Database is a fully managed relational database — SQL Server compatibility, ACID transactions, foreign keys, and complex joins. Best for structured relational workloads. Cosmos DB is a globally distributed, multi-model database with single-digit millisecond latency at any scale — it supports multiple APIs (SQL/Core, MongoDB, Cassandra, Table, Gremlin). Best for globally distributed workloads requiring low latency and flexible schemas. Azure Table Storage is simple, cheap NoSQL key-value storage — it is less capable than Cosmos DB but lower cost for workloads that do not need global distribution or multiple APIs. When cost is the constraint and the data model is simple, Table Storage. When global distribution and SLA-backed latency matter, Cosmos DB.

Quick check

Is this a relational workload with SQL and joins (Azure SQL), a globally distributed low-latency flexible-schema workload (Cosmos DB), or simple key-value data at minimal cost (Table Storage)?

Why it looks right

Cosmos DB is the modern answer for NoSQL and candidates apply it universally. Table Storage is the correct lower-cost answer for simple key-value workloads where Cosmos DB capabilities are not required.

Azure VPN GatewayAzure ExpressRoute
#6

Encrypted tunnel over the internet vs. dedicated private circuit

Both connect on-premises networks to Azure, so candidates apply ExpressRoute as the "better" option without checking cost or setup time.

Deciding signal

VPN Gateway creates an encrypted IPsec/IKE tunnel from an on-premises device to Azure over the public internet. It is fast to set up, cost-effective, and suitable when consistent high bandwidth and dedicated latency are not critical requirements. ExpressRoute provides a private dedicated circuit from your data center to Azure through a connectivity provider — it bypasses the public internet, offers guaranteed bandwidth SLAs, and provides lower and more consistent latency. ExpressRoute takes weeks to provision and costs significantly more. When the scenario describes predictable latency, high bandwidth, and security requirements for a large enterprise connection, ExpressRoute. When it describes a faster, lower-cost hybrid connection, VPN Gateway.

Quick check

Is this an encrypted internet-based connection that is fast to provision (VPN Gateway), or a private dedicated circuit with guaranteed bandwidth and consistent latency (ExpressRoute)?

Why it looks right

ExpressRoute is the premium option and candidates associate it with all enterprise hybrid connectivity. VPN Gateway is the correct answer when cost, setup speed, or internet-based encryption is what the scenario describes.

Azure CDNAzure Front DoorAzure Traffic Manager
#7

Edge content caching vs. global HTTP load balancing vs. DNS-based global routing

All three improve performance for globally distributed users, so candidates apply Front Door as the default "global" answer.

Deciding signal

Azure CDN caches static content (images, CSS, JavaScript, video) at edge POPs, reducing origin load and improving download speed for cacheable assets. It does not provide global load balancing of dynamic traffic. Azure Front Door is a global HTTP/HTTPS load balancer with routing rules, WAF integration, SSL offload, and intelligent routing across backend origins — it handles both static and dynamic traffic with URL-based routing. Traffic Manager is a DNS-based traffic routing service — it does not proxy traffic (no content delivery or HTTP termination), it just resolves DNS to different IP addresses based on routing policies (performance, weighted, geographic, priority). When the scenario involves HTTP routing with WAF and health-based failover, Front Door. When it involves DNS-based routing to regional endpoints without HTTP inspection, Traffic Manager.

Quick check

Is this caching static content at edge (CDN), globally routing and load balancing HTTP/HTTPS with WAF (Front Door), or DNS-based routing to regional endpoints without HTTP proxy (Traffic Manager)?

Why it looks right

Front Door and Traffic Manager both "route global traffic" and candidates conflate them. Traffic Manager only resolves DNS — it cannot inspect HTTP content, integrate WAF, or cache. Front Door does all of those.

Microsoft Defender for CloudMicrosoft SentinelMicrosoft Entra ID Protection
#8

Cloud workload protection vs. SIEM/SOAR vs. identity risk detection

All three detect and respond to security threats, so candidates apply Defender for Cloud to all security posture questions.

Deciding signal

Defender for Cloud (formerly Azure Security Center) assesses and protects Azure workloads — VMs, databases, containers, storage. It provides a secure score, hardening recommendations, and threat detection for Azure resources. Microsoft Sentinel is a cloud-native SIEM and SOAR — it aggregates logs from Azure, on-premises, and third-party sources, applies ML analytics to detect threats, and provides playbooks for automated response. Entra ID Protection detects risky sign-ins and compromised user accounts using ML on identity signals. When the scenario involves protecting Azure resources and improving their security configuration, Defender for Cloud. When it involves aggregating logs for threat hunting and incident response across an entire environment, Sentinel. When it involves detecting compromised user accounts or risky login behavior, Entra ID Protection.

Quick check

Is this assessing and protecting Azure workload configurations (Defender for Cloud), aggregating security logs for threat hunting and automated response (Sentinel), or detecting risky sign-ins and compromised identities (Entra ID Protection)?

Why it looks right

Defender for Cloud is the familiar security posture service. Sentinel is correct when the scenario involves SIEM capabilities — log aggregation, custom detection rules, and incident investigation across a broad environment.

Azure PolicyAzure RBAC
#9

Resource compliance enforcement vs. identity permission assignment

Both control what happens in Azure subscriptions, so candidates conflate compliance guardrails with access permissions.

Deciding signal

Azure RBAC assigns permissions to identities — it controls what actions a user, group, or service principal can perform on resources. It answers "who can do what." Azure Policy evaluates resource properties against defined rules and either audits or enforces compliance — it answers "are resources configured correctly?" For example, RBAC might allow a user to create storage accounts; Azure Policy might enforce that all storage accounts must be created with HTTPS-only access enabled. Policy can deny non-compliant resource creation or remediate existing resources. They are complementary: RBAC controls user actions; Policy controls resource configuration regardless of who created them.

Quick check

Is the requirement to control who can perform actions on resources (RBAC), or to enforce that resources meet specific configuration standards regardless of who creates them (Azure Policy)?

Why it looks right

RBAC is the familiar access control tool. Azure Policy is the correct answer when the scenario describes enforcing a resource configuration rule across a subscription or management group — a constraint on the resource, not on the user.

Azure Reserved InstancesAzure Spot VMsAzure Hybrid Benefit
#10

Long-term commitment discount vs. interruptible capacity discount vs. license portability

All three reduce Azure VM costs, so candidates apply Reserved Instances to all cost-optimization scenarios.

Deciding signal

Reserved Instances provide up to 72% savings versus pay-as-you-go in exchange for a 1- or 3-year commitment to a specific VM size in a specific region. Best for stable, predictable workloads. Spot VMs offer deep discounts (up to 90%) on unused Azure capacity, but Azure can evict them with 30 seconds notice when capacity is needed. Best for interruptible batch jobs or stateless workloads. Azure Hybrid Benefit allows using existing on-premises Windows Server or SQL Server licenses with Software Assurance on Azure VMs, reducing the cost of the licensing component. These are not mutually exclusive — Hybrid Benefit and Reserved Instances can be combined.

Quick check

Is this a stable workload needing a long-term commitment discount (Reserved Instances), an interruptible workload that can tolerate eviction (Spot VMs), or reusing existing on-premises Windows/SQL licenses (Hybrid Benefit)?

Why it looks right

Reserved Instances are the most visible cost-reduction tool. Spot VMs are the correct answer when the scenario describes workloads that can be interrupted — a signal that candidates sometimes overlook in favor of Reserved Instances.

Train these confusions, not just read them

10 AZ-900 questions. Pattern-tagged with trap analysis. Free, no signup required.

Start AZ-900 Mini-Trainer →