CISSP domain - 13% of the exam

Security Architecture and Engineering

Security Architecture and Engineering is 13% of the CISSP exam. These are the objectives it covers, each with practice questions, with every answer explained.

The domain in numbers

39
Practice questions
10
Objectives
4
Easy
19
Medium
16
Hard

Where people go wrong

  • Attaching a property to the wrong formal model

    The wrong answers attach a property or construct to the wrong formal model. Bell-LaPadula enforces confidentiality with no read up and no write down, while the simple integrity and star integrity properties belong to Biba. Clark-Wilson routes every change through a transformation procedure, and integrity verification procedures only check state. Brewer-Nash covers conflicts of interest between client datasets. A state machine model proves every reachable state secure, and an information flow model constrains flows between classes.

    See it in a sample question
  • Swapping the jobs of TPM, enclave, ASLR and NX

    The wrong answers swap the job of one hardware or memory protection for another. A TPM is a co-processor for key storage, sealing and platform measurement, not bulk encryption or runtime isolation. A TEE runs trusted code in an isolated context alongside the rich operating system, and a secure enclave keeps its memory unreadable even to the kernel and hypervisor. ASLR randomises where code and data sit, while the NX bit marks data pages as non-executable. DEP and sealing do not hide memory from a privileged user.

    See it in a sample question
  • Naming the wrong attack for the behaviour described

    The wrong answers name the wrong attack for the behaviour described. Pass-the-hash replays a captured NTLM hash as the credential without recovering the password, so it is neither brute force nor a birthday attack. Side-channel attacks read timing, power or electromagnetic leakage from an implementation. Fault injection perturbs one cryptographic operation so the faulty output reveals key bits, and key length does not defend against it. Ransomware encrypts files with a symmetric key and protects that key under the attacker's public key.

    See it in a sample question
  • Hardening beside a shared dependency instead of removing it

    The wrong answers harden something near the weakness while the shared dependency stays in place. Default provider keys, a web application firewall or stronger authentication leave production and development in one account. Pod resource limits and credential rotation leave pods holding the node's cloud role. Credential changes and perimeter detection leave the pumps on a flat clinical VLAN. A longer battery leaves a common automatic transfer switch and day-tank. The correct answers remove it: separate accounts, scoped workload identities, a segmented zone for the pumps, or a second transfer switch.

    See it in a sample question
  • Leading facility design with hardening or uniform zoning

    The wrong answers lean on hardening or uniform treatment where layout should lead. Bolted furniture and tougher perimeters are target hardening, which CPTED layers on after sightlines, access control and territorial reinforcement. Siting the server room on the public face, on an exterior wall, or in one flat zone exposes it. The correct answers place the teller area and entrance for clear sightlines and arrange spaces in concentric zones, with the server room in the most protected interior.

    See it in a sample question

Objectives in this domain

Sample questions from this domain

Free sampleSecurity Architecture and Engineeringhard

A defence research lab is selecting a formal security model for a multi-level workstation that will hold material at Confidential, Secret, and Top Secret. Auditors have stressed that an analyst cleared to Secret must never see Top Secret intelligence on screen, but may be required to file routine status notes into a Top Secret incident log. Which statement most accurately characterises how the Bell-LaPadula model handles this requirement?

  • ABell-LaPadula enforces confidentiality through the simple security property (no read up) and the star property (no write down), which together permit the Secret analyst to append to a Top Secret log without exposing Top Secret content. Correct
  • BBell-LaPadula relies on the simple integrity property and the star integrity property, which would prevent the Secret analyst from contributing to a Top Secret log because writing up could corrupt higher-assurance data.
  • CBell-LaPadula uses a strong tranquillity principle that freezes labels for the duration of a session, which would force the Secret analyst to be re-cleared to Top Secret before any append could be performed.
  • DBell-LaPadula applies the discretionary security property as its primary control, which would let the data owner grant the Secret analyst a read-up exception on a per-file basis for the Top Secret log.
Identify how Bell-LaPadula's simple security and star properties protect confidentiality while permitting writes to higher classifications. Bell-LaPadula is a confidentiality-only model built on lattice-based mandatory access control. The simple security property forbids a subject from reading an object whose classification dominates the subject's clearance, while the star property forbids writing to an object whose classification is dominated by the subject's clearance. Together these allow a lower-cleared subject to append blindly to a higher-classified object, which is the exact behaviour described.

Why A is correct: Correct. The simple security property prevents reading objects at a higher classification, and the star property prevents writing to a lower classification, so a Secret subject can write up to a Top Secret object but cannot read its contents, matching the lab's requirement precisely.

Why B is wrong: Tempting because the property names sound parallel, but those are Biba integrity properties, not Bell-LaPadula confidentiality properties; Bell-LaPadula is silent on integrity and would not block a write up on integrity grounds.

Why C is wrong: Strong tranquillity is a real Bell-LaPadula concept, but it constrains label changes on subjects and objects, not the legitimacy of a write-up; a Secret subject may still write to a Top Secret object without any clearance change.

Why D is wrong: The discretionary security property does exist in Bell-LaPadula as a need-to-know overlay, but it cannot override the mandatory simple security property; no discretionary grant can authorise reading up to Top Secret.

Free sampleSecurity Architecture and Engineeringmedium

A platform engineer is comparing a Trusted Execution Environment (TEE) such as ARM TrustZone with a discrete TPM on the same mobile device. Which statement best captures the primary distinction between the two capabilities?

  • AA TEE protects keys at rest in tamper-resistant storage, while a TPM provides an isolated processor mode for executing sensitive code at runtime.
  • BA TEE provides an isolated execution context for trusted code and data alongside the rich operating system, while a TPM is a co-processor focused on key storage, sealing, and platform measurement. Correct
  • CA TEE is a software sandbox enforced by the hypervisor for guest workloads, while a TPM is a firmware feature exposed through ACPI tables to the operating system loader.
  • DA TEE encrypts the file system using device-bound keys, while a TPM signs application binaries during installation to enforce code provenance.
Distinguish a TEE as an isolated runtime execution environment from a TPM as a key-protection and measurement co-processor. A Trusted Execution Environment leverages CPU features such as ARM TrustZone or processor enclave instructions to create a secure world in which trusted code can process secrets without exposure to the rich operating system. A TPM, by contrast, is a co-processor that stores cryptographic keys, performs limited primitives, and maintains integrity measurements in Platform Configuration Registers. They address different threats: TEE addresses runtime confidentiality of code and data, TPM addresses key protection and platform attestation.

Why A is wrong: This reverses the two roles. TPMs are oriented toward key storage and platform measurement, while a TEE provides the isolated execution context. A candidate who memorises the labels without the function can swap them.

Why B is correct: TEEs partition the main CPU into a secure world that runs trusted applications separated from the normal world, whereas TPMs are passive co-processors holding keys and integrity measurements. The two are complementary rather than equivalent.

Why C is wrong: A TEE is not a hypervisor sandbox; it relies on CPU mode separation enforced by hardware. Describing the TPM purely as an ACPI feature understates its role as a tamper-resistant co-processor with its own command interface.

Why D is wrong: Neither description is accurate. File system encryption is performed by operating system or storage components, and code signing is performed by signing authorities, not by the TPM during installation.

Free sampleSecurity Architecture and Engineeringhard

A security architect is comparing classes of cryptanalytic attack for a threat model covering a smart card that performs RSA signatures. Which statement BEST distinguishes a side-channel attack from a traditional cryptanalytic attack?

  • AA side-channel attack derives key material from unintended physical leakage such as timing, power, or electromagnetic emissions during cryptographic operations. Correct
  • BA side-channel attack exploits weaknesses in the mathematical structure of the cipher, while a traditional cryptanalytic attack targets the physical implementation.
  • CA side-channel attack requires the attacker to recover the plaintext from intercepted ciphertext without any access to the device performing the encryption.
  • DA side-channel attack succeeds only when the underlying algorithm has a known mathematical break that reduces its effective key length.
Differentiate side-channel attacks, which exploit physical leakage of an implementation, from classical cryptanalysis targeting the algorithm itself. Side-channel attacks exploit information leaked during the physical execution of a cryptographic operation, such as power draw, timing, cache behaviour, or electromagnetic emissions. The cipher can be mathematically sound and still leak key bits if the implementation is not constant-time, blinded, or shielded, which is why countermeasures focus on the implementation rather than the algorithm.

Why A is correct: Side-channel attacks treat the device as a leaky black box and recover secrets from observable physical correlates of the computation rather than by analysing ciphertext or known plaintext.

Why B is wrong: This reverses the definitions. Mathematical weaknesses in the cipher are what classical cryptanalysis (linear, differential, algebraic) targets; side-channel work deliberately ignores the maths and attacks the implementation.

Why C is wrong: Ciphertext-only recovery describes a classical attack model. Side-channel work almost always requires proximity or oracle-like access to the device while it operates on a secret.

Why D is wrong: Side-channel attacks routinely defeat algorithms that have no known mathematical weakness, including AES and RSA, because they target the implementation rather than the algorithm.

Free sampleSecurity Architecture and Engineeringhard

A retailer is migrating its booking platform to a public IaaS provider and the architecture board is reviewing the design. The cloud team proposes that all workloads run in a single shared account, that production secrets be stored in the same key vault as development, and that the provider's default identity federation be enabled for every staff role. The CISO wants the BEST first change to reduce the architectural blast radius before any workloads ship. Which change should the board mandate?

  • ASeparate production, non-production, and security tooling into distinct cloud accounts with their own identity boundaries and dedicated key management. Correct
  • BEnable provider-managed encryption at rest with default keys across every storage service in the single shared account.
  • CDeploy a web application firewall in front of every public endpoint and route all developer traffic through it.
  • DRequire multi-factor authentication for all developer console logins and rotate static access keys every ninety days.
Recognise that separating cloud accounts is the primary architectural control for limiting blast radius in IaaS designs. In public cloud IaaS, the account (or subscription/project) is the strongest available isolation boundary for identity, networking, and key management. Co-locating production and non-production in one account means a single over-privileged role, leaked key, or misconfigured trust policy can traverse environments. Establishing distinct accounts per environment, with their own identity stores and key management, structurally limits how far any one compromise can spread, which the other options cannot do because they harden individual controls inside an already-fused trust boundary.

Why A is correct: Account-level separation is the strongest control plane boundary in IaaS, because a compromise of one workload cannot pivot to another account's identities, keys, or networks without crossing an explicit trust. It addresses the architectural blast radius before the workloads are even built, which is exactly what the board is being asked to mandate.

Why B is wrong: Default provider keys do raise the bar against raw disk theft, but they do not stop a compromised identity in the shared account from reading the data, and they leave the production and development blast radius fused together. It is a useful baseline control but not the architectural fix the question demands.

Why C is wrong: A web application firewall reduces certain application-layer attacks but does nothing about shared accounts, shared key vaults, or over-broad federation, so the structural weakness in the design remains. It is a perimeter control bolted onto a flawed foundation.

Why D is wrong: Stronger authentication and key rotation harden individual identities, yet they leave production and development sharing one trust boundary and one key vault. A single compromised privileged session would still reach production secrets, so this hardens users without fixing the architecture.

Free sampleSecurity Architecture and Engineeringeasy

A logistics company is fitting out a new head office with a customer reception, open-plan staff floors, a finance team area, and a small server room. The facility designer asks the security architect how to organise these spaces. What is the BEST guiding principle for the layout?

  • APlace the server room next to reception so visitors can see that security is taken seriously.
  • BArrange spaces in concentric zones, with public areas at the perimeter and the server room in the most protected interior zone. Correct
  • CTreat all internal spaces as equally sensitive and apply the same access controls everywhere to keep the design simple.
  • DLocate the finance area and the server room on an exterior wall so they have natural light and clear ventilation paths.
Apply layered concentric zoning so the most sensitive facility spaces sit in the deepest, best-protected interior zone. Secure facility design uses progressive zones of trust. Each successive zone, from public to restricted, adds controls such as access cards, mantraps, and monitoring. Locating critical assets such as a server room in the innermost zone forces an attacker through multiple controls and shortens detection time.

Why A is wrong: Putting the server room on the public face of the building advertises a high-value target and shortens the path for an intruder. Visible commitment to security should come from policy and signage, not from siting sensitive infrastructure in public view.

Why B is correct: Layered or concentric zoning matches defence in depth for facilities. Public reception sits on the outer ring, staff areas inside that, and the highest value asset, the server room, sits in the innermost zone with the strongest controls and fewest entry paths.

Why C is wrong: Flat zoning is easy to manage but wastes control on low-value areas and under-protects high-value ones. Site and facility design expects controls to be proportionate to the sensitivity of each space rather than uniform across the building.

Why D is wrong: Exterior walls expose sensitive areas to direct attack through windows, vehicles, and utility runs. Light and ventilation are real needs, but they are met by mechanical systems rather than by placing high-value spaces on the building skin.

Other domains in this exam

See also the CISSP cert hub, the study guide, and the cheat sheet.

Examworthy is not affiliated with or endorsed by ISC2. Original, blueprint-aligned practice material only.