SC-100 domain - 27% of the exam

Design Security Solutions for Infrastructure

Design Security Solutions for Infrastructure is 27% of the SC-100 exam. These are the objectives it covers, each with practice questions, with every answer explained.

The domain in numbers

85
Practice questions
4
Objectives
42
Medium
43
Hard

Where people go wrong

  • Putting agent-based Defender products on agentless OT devices

    The wrong answers offer Defender for Servers, Defender for Endpoint or a cloud-connected sensor for plant controllers, embedded field devices and unmanaged IoT. Those need an installed agent or a cloud path, which constrained operational technology devices cannot provide. Microsoft Defender for IoT discovers assets passively from mirrored network traffic and baselines industrial protocols without touching the devices. For an isolated network with no internet route, locally managed sensors and an on-premises management console keep the data inside.

    See it in a sample question
  • Reaching for a VPN or web gateway where per-application access is needed

    The wrong options for reaching internal systems are point-to-site or site-to-site VPN, Azure Bastion, Application Gateway, Microsoft Entra Internet Access and application proxy. A VPN grants network-layer reachability to whole subnets, Application Gateway opens a public listener, Internet Access covers internet and software-as-a-service traffic, and application proxy publishes only web applications. Microsoft Entra Private Access publishes each internal application, including non-HTTP TCP ones, through an outbound connector with Conditional Access enforced per application.

    See it in a sample question
  • Anchoring trust to a network address

    Named locations, IP allow lists and office address ranges are offered as proof that a session is trustworthy. An address does not show identity or device state. It fails for remote staff and can be satisfied by routing through a known egress. The correct answers drop network location as a basis for trust. The Microsoft Entra Internet Access compliant-network check in Conditional Access requires traffic to arrive through the secure edge, and per-session brokering through Private Access verifies user and device every time.

    See it in a sample question
  • Confusing posture management with runtime threat detection

    Posture tooling and runtime detection get substituted for each other. Defender for Cloud posture recommendations, Azure Policy for Kubernetes and Intune baselines assess configuration, so none of them alerts on a running crypto-miner, a reverse shell or tampering with critical files. The workload protection plans do that: Defender for Servers, Defender for Containers, and file integrity monitoring in Defender for Servers Plan 2. The reverse mix-up appears too, where Defender for Servers is offered for attack path analysis, which needs the Defender CSPM plan.

    See it in a sample question
  • Offering Sentinel where a posture or protection plan is needed

    Microsoft Sentinel appears as the wrong option in questions about multicloud server protection, posture assessment, bespoke controls and container images. It is a SIEM that correlates collected telemetry, so it does not assess resource configuration against a benchmark, scan container image layers, or protect server workloads. Defender for Cloud does those jobs: multicloud posture assessment against the Microsoft cloud security benchmark, Defender for Servers protection across Azure and Amazon Web Services, custom security standards for bespoke controls, and Defender for Containers image scanning.

    See it in a sample question

Objectives in this domain

What to study

The official documents this domain's practice questions cite most.

Sample questions from this domain

Free sampleDesign Security Solutions for Infrastructuremedium

A manufacturer runs a plant floor of programmable logic controllers and human-machine interfaces that cannot accept an installed agent and must not be rebooted for monitoring. The security architect needs a design that passively discovers these operational technology assets, builds a behavioural baseline of industrial protocols, and alerts on anomalies without disrupting production. Which Microsoft capability fits this operational technology monitoring requirement?

  • AMicrosoft Defender for Servers, deployed through Microsoft Defender for Cloud to install endpoint sensors on the plant controllers so that industrial protocol anomalies are baselined and alerted without rebooting the devices.
  • BMicrosoft Defender for IoT, deployed with passive network sensors that mirror plant traffic to discover operational technology assets, baseline industrial protocols, and alert on anomalies without touching the devices. Correct
  • CMicrosoft Defender for Endpoint, onboarded to each programmable logic controller and human-machine interface so that the operational technology estate gains behavioural baselining of industrial protocols and anomaly alerts.
  • DMicrosoft Defender for Cloud Apps, configured with session policies that inspect the plant control traffic so that operational technology protocol anomalies are baselined and alerted without an agent on the controllers.
Select Microsoft Defender for IoT for agentless passive monitoring, asset discovery, and anomaly detection across operational technology and industrial control networks. Operational technology devices such as PLCs and HMIs rarely tolerate installed agents or reboots, so the design must monitor them passively. Microsoft Defender for IoT places network sensors that mirror traffic and perform deep packet inspection of industrial protocols, discovering assets and learning a behavioural baseline to alert on anomalies without ever touching the controllers, which agent-based endpoint products cannot do here.

Why A is wrong: Defender for Servers is tempting as the workload protection plane, but it protects Windows and Linux server operating systems with an installed sensor, which the agentless plant controllers in this scenario cannot host.

Why B is correct: Defender for IoT uses agentless passive network monitoring and deep packet inspection of industrial protocols to discover and baseline operational technology assets and flag anomalies, which exactly matches a non-intrusive plant-floor requirement.

Why C is wrong: Defender for Endpoint appeals because it is the core endpoint protection product, but it requires an onboarded operating system agent that embedded industrial controllers do not support, so it cannot passively monitor this operational technology.

Why D is wrong: Defender for Cloud Apps is tempting as an agentless inspection layer, but it is a cloud access security broker for SaaS and web app sessions, not a sensor that understands industrial control protocols on a plant network.

Free sampleDesign Security Solutions for Infrastructurehard

An organisation wants to retire its always-on corporate VPN that grants connected devices broad network-layer reachability to internal subnets. The design must instead let users reach only the specific private applications they are authorised for, with Conditional Access evaluated per application and no inbound ports opened on the corporate network. Which Microsoft capability should the architect select?

  • AMicrosoft Entra Internet Access, configured to tunnel all outbound user traffic through the cloud edge so the legacy VPN is no longer needed for any destination.
  • BMicrosoft Entra Private Access, publishing each internal application so authenticated users reach only authorised apps through an outbound connector with Conditional Access enforced per application. Correct
  • CAn Azure point-to-site VPN gateway with split tunnelling, scoping the advertised routes so that only the subnets hosting the line-of-business applications are reachable by connected clients.
  • DAzure Bastion in front of the application servers so administrators and users connect over a hardened browser session without exposing the servers to the public internet.
Select Microsoft Entra Private Access to replace broad VPN reachability with least-privilege, per-application Conditional Access for internal apps. Microsoft Entra Private Access is the private-access pillar of Microsoft Entra Security Service Edge. It publishes internal applications through outbound connectors so no inbound ports are opened, and access is brokered per application with Conditional Access, which removes the broad network-layer reachability that a traditional always-on VPN grants.

Why A is wrong: Microsoft Entra Internet Access is attractive because it also replaces a VPN tunnel, but it secures access to internet and software-as-a-service destinations rather than publishing internal private applications, so it cannot deliver per-application access to on-premises subnets.

Why B is correct: Microsoft Entra Private Access publishes individual internal applications through an outbound-only connector and enforces Conditional Access per application, giving least-privilege access without broad subnet reachability or inbound ports, which is precisely the requirement.

Why C is wrong: Route scoping on a point-to-site gateway looks like least privilege, but it still grants network-layer reachability to whole subnets and lacks per-application Conditional Access, so a compromised client can still move laterally within the advertised ranges.

Why D is wrong: Azure Bastion is tempting because it avoids public exposure, but it provides secure remote desktop and shell access to virtual machines rather than general per-application access for a workforce, so it does not satisfy a broad authorised-application requirement.

Free sampleDesign Security Solutions for Infrastructurehard

An architect wants Conditional Access policies for a sensitive software-as-a-service application to be able to require that the user's traffic egresses through the organisation's secure cloud edge, so that source-anchored network conditions cannot be spoofed by simply browsing from an arbitrary address. Which Microsoft capability provides the network signal that Conditional Access can require for this purpose?

  • AConfigure trusted named locations in Conditional Access using the published egress IP addresses of the corporate datacentre, and require sign-ins to the application to originate from those trusted locations.
  • BEnable continuous access evaluation on the application so that token revocation happens in near real time whenever the user's network address changes during an active session.
  • CAdopt Microsoft Entra Internet Access so that the user's traffic is tunnelled through the secure cloud edge, and require the compliant-network check in Conditional Access for the application. Correct
  • DUse Azure Firewall with a static public IP for outbound traffic and add that address to the application's allow list, so only flows that traversed the firewall reach the software-as-a-service tenant.
Use the Microsoft Entra Internet Access compliant-network check in Conditional Access to source-anchor access without relying on spoofable IP locations. Microsoft Entra Internet Access establishes a tunnel through the Microsoft secure cloud edge and signals a compliant-network condition that Conditional Access policies can require. Because the condition reflects verified passage through the edge rather than a declared source address, it resists the IP-spoofing weakness of named locations and provides a robust, identity-integrated network signal.

Why A is wrong: Trusted named locations are the traditional way to assert network context and are tempting, but they bind trust to static IP ranges that break for remote users and can be approximated by routing through a known address, so they do not provide a tamper-resistant edge signal.

Why B is wrong: Continuous access evaluation reacts quickly to changes such as network moves and is valuable, but it evaluates and revokes existing tokens rather than supplying a network-egress signal that a policy can require at grant time, so it does not satisfy the source-anchoring requirement.

Why C is correct: Microsoft Entra Internet Access tunnels user traffic through the secure cloud edge and emits a compliant-network signal that Conditional Access can require, which anchors access to verified passage through the organisation's edge rather than to a spoofable source IP, meeting the requirement.

Why D is wrong: Pinning outbound traffic to a firewall public IP and allow-listing it at the application sounds like source anchoring, but it enforces at the application rather than through Conditional Access and is still an IP allow list that an attacker reaching that egress could ride, so it is not the identity-plane signal sought.

Free sampleDesign Security Solutions for Infrastructurehard

A bank already uses Microsoft Defender for Cloud for posture recommendations across its Azure and Arc-enabled servers, but a recent incident showed it had no runtime detection of a crypto-mining process and a suspicious shell launched on those servers. The architect must add the design capability that detects and alerts on active threats executing on the workloads themselves. Which addition meets this requirement?

  • ATighten the foundational cloud security posture management recommendations and raise the Secure Score target so that the strengthened configuration baseline detects and alerts on active threats running on the servers.
  • BEnable the Microsoft Defender for Servers workload protection plan in Defender for Cloud so that runtime threat detection alerts on behaviours such as crypto-mining and suspicious shell execution on the protected machines. Correct
  • CSubscribe the servers to Microsoft Defender External Attack Surface Management so its outside-in monitoring detects and alerts on the crypto-mining and suspicious shell activity occurring on the workloads.
  • DApply an Azure Policy initiative that audits running processes on the servers so that non-compliant workloads are flagged when crypto-mining or a suspicious shell is executed on them.
Add the Microsoft Defender for Servers workload protection plan to gain runtime threat detection beyond configuration posture management. Posture management assesses how resources are configured, but it cannot see malicious behaviour at runtime. The Defender for Servers cloud workload protection plan adds behavioural threat detection on the machines, alerting on activity such as crypto-mining and suspicious shells, which posture scoring, perimeter discovery, and policy auditing cannot detect.

Why A is wrong: Hardening posture recommendations is tempting because it reduces attack surface, but cloud security posture management assesses configuration and cannot detect a malicious process executing at runtime, which is the gap the incident exposed.

Why B is correct: Defender for Servers is the cloud workload protection plan that adds runtime threat detection and behavioural alerting on machines, which is precisely what detects the crypto-mining and suspicious shell activity the incident missed.

Why C is wrong: External Attack Surface Management is appealing for continuous monitoring, but it discovers and assesses internet-facing assets from outside and has no visibility into processes executing inside a server's operating system.

Why D is wrong: Azure Policy is tempting because it can audit guest configuration, but it evaluates declarative configuration state on a schedule and does not provide the real-time behavioural threat detection needed to alert on active malicious processes.

Free sampleDesign Security Solutions for Infrastructuremedium

An enterprise running Windows and Linux virtual machines in Azure and on Amazon Web Services wants threat detection, vulnerability assessment, and security recommendations for those server workloads delivered through a single multicloud plane that also feeds its posture scoring. Which design should the architect specify to protect these server endpoints consistently across both clouds?

  • ASpecify Microsoft Sentinel as the multicloud plane, ingesting server logs from Azure and Amazon Web Services so that one analytics workspace delivers threat detection, vulnerability assessment, and security recommendations for the server workloads.
  • BSpecify Microsoft Defender for Servers through Microsoft Defender for Cloud, connecting Azure and Amazon Web Services so that one plane delivers threat detection, vulnerability assessment, and recommendations that feed posture scoring. Correct
  • CSpecify Microsoft Intune as the multicloud plane, enrolling the Azure and Amazon Web Services virtual machines so that one console delivers threat detection, vulnerability assessment, and security recommendations for the server workloads.
  • DSpecify Microsoft Entra ID Protection as the multicloud plane, evaluating the Azure and Amazon Web Services server identities so that one service delivers threat detection, vulnerability assessment, and security recommendations for the server workloads.
Select Microsoft Defender for Servers via Defender for Cloud to deliver multicloud server workload protection, vulnerability assessment, and posture scoring. Consistent server endpoint protection across Azure and AWS requires a cloud workload protection plane, not a SIEM or an identity engine. Microsoft Defender for Servers, enabled through Microsoft Defender for Cloud with multicloud connectors, brings threat detection, integrated vulnerability assessment, and hardening recommendations to Windows and Linux virtual machines in both clouds while contributing to secure score for unified posture management.

Why A is wrong: Sentinel is tempting because it spans multiple clouds, but it is a SIEM and SOAR layer for detection and response across logs rather than the workload protection plane that performs vulnerability assessment and posture recommendations for server endpoints.

Why B is correct: Defender for Servers, delivered via Defender for Cloud with multicloud connectors, protects Windows and Linux servers across Azure and AWS with threat detection and vulnerability assessment while feeding secure score, which is exactly the single-plane server protection required.

Why C is wrong: Intune is appealing as a unified management console, but it manages and configures client and mobile endpoints rather than providing cloud workload threat detection and vulnerability assessment for multicloud server virtual machines.

Why D is wrong: Entra ID Protection is tempting as a risk engine, but it detects identity risk and drives risk-based access policy rather than performing server workload threat detection, vulnerability assessment, or posture recommendations across clouds.

Other domains in this exam

See also the SC-100 cert hub, the study guide, and the cheat sheet.

Examworthy is not affiliated with or endorsed by Microsoft. Original, blueprint-aligned practice material only.