Cloud Platform and Infrastructure Security

A cloud security team is documenting BC/DR targets for a hosted order-processing service. They need to state the maximum amount of data, measured as a period of time, that the business can afford to lose if the service fails and must be recovered from the most recent backup. Which metric expresses this?

• ARecovery time objective (RTO), the maximum tolerable period during which data may be lost before the service is restored to operation.
• BRecovery point objective (RPO), the maximum tolerable period during which data may be lost, which sets how frequently backups or replication must occur.check_circle Correct
• CRecovery service level (RSL), the percentage of computing capacity that must be available during recovery operations.
• DMaximum tolerable downtime (MTD), the total outage the business can absorb before its viability is threatened.

Why A is wrong: RTO is a real metric and is tempting because it is a time-based recovery target, but it measures how long restoration may take, not how much data may be lost.

Why B is correct: RPO is defined as the maximum acceptable age of the data at recovery, expressed as a time interval, so it directly fixes the backup or replication frequency.

Why C is wrong: RSL is a genuine metric describing the proportion of normal capacity delivered during recovery, but it addresses performance during degraded operation, not tolerated data loss.

Why D is wrong: MTD is real and bounds RTO, but it caps the length of the outage rather than the quantity of data lost, so it does not answer the question.

Under a serverless (function-as-a-service) compute model, which security responsibility remains with the customer rather than shifting to the cloud provider?

• ASecuring the function code, its dependencies and the identity permissions granted to each function.check_circle Correct
• BPatching and hardening the operating system that hosts each function invocation.
• CMaintaining the physical security of the data centres where the function runtime executes.
• DScaling the underlying execution fleet up and down to match the volume of incoming invocations.

Why A is correct: Even when the provider runs the platform, the customer still writes the function code, chooses its dependencies and defines the permissions it holds, so application and identity security stays with the customer.

Why B is wrong: This is tempting because patching matters everywhere, but in serverless the provider manages the underlying operating system and runtime host, so this responsibility shifts away from the customer.

Why C is wrong: Physical data-centre security is always the provider's responsibility across every cloud service model, so it is never the customer's task under serverless.

Why D is wrong: Automatic scaling of the execution infrastructure is a core provider-managed feature of serverless, so the customer does not own it.

Two data centre engineers debate terminology while reviewing redundancy diagrams. One asks which statement correctly describes an N+1 power configuration. Which statement is TRUE?

• AIt provides the exact capacity needed to run the load with no spare capacity component available.
• BIt provides two fully independent power systems, each able to carry the entire load alone.
• CIt provides one spare capacity component in addition to the number required to carry the load.check_circle Correct
• DIt provides double the required capacity plus one extra unit on every distribution path.

Why A is wrong: That description is plain N, the baseline capacity with no redundancy; N+1 by definition adds a spare beyond the required amount, so this omits the redundant component that makes it N+1.

Why B is wrong: Two independent systems each sized for the full load describes 2N, a distribution-path level of redundancy; N+1 adds only a single spare component, so this overstates the redundancy involved.

Why C is correct: N+1 means the number of components needed to serve the load, N, plus one additional unit, so a single component can fail or be serviced while the remaining units still carry the full load.

Why D is wrong: Doubling capacity and adding a spare describes 2N+1, a higher redundancy class; conflating it with N+1 confuses component-level sparing with fully mirrored, independently pathed systems.

A data centre designer needs continuous power for the IT load during the seconds between a mains utility failure and the moment the standby diesel generators reach full output. Which component is specifically intended to bridge that gap?

• AThe power distribution units feeding each rack from the floor busway.
• BThe uninterruptible power supply drawing on its stored energy reserve.check_circle Correct
• CThe automatic transfer switch selecting between utility and generator feeds.
• DThe standby generator's own fuel supply held in the day tank.

Why A is wrong: Power distribution units divide and route conditioned power to racks, but they store no energy of their own; if the source loses power they pass the outage straight through, so they cannot bridge the transfer gap.

Why B is correct: The uninterruptible power supply holds a battery or flywheel reserve that instantly carries the IT load when mains fails, keeping equipment running for the seconds until the generators stabilise and take over.

Why C is wrong: The automatic transfer switch decides which source feeds the load and initiates the generator start, but it stores no energy; the load would still drop during the crank and warm-up delay without a reserve behind it.

Why D is wrong: The day tank ensures the generator can run once started, but the generator itself needs seconds to start and reach full output, so the fuel supply does not cover the initial gap the question describes.

A logistics company runs its fleet-tracking application on virtual machines in a public-cloud IaaS deployment. The data processed is classified as internal. The security team wants to reduce the risk that a compromise of one tenant's workload on the same physical host could reach their virtual machines. Which control most directly addresses this concern at the correct layer?

• ADeploy a host-based intrusion detection agent inside each of the company's virtual machines
• BApply network micro-segmentation between the company's own application tiers
• CEnforce strong isolation and patching of the type 1 hypervisor that separates tenant virtual machines on the shared hostcheck_circle Correct
• DEncrypt the application data at rest using customer-managed keys

Why A is wrong: A host agent gives useful visibility inside the guest, but it sits above the isolation boundary and cannot prevent or detect an escape happening at the hypervisor beneath it, so it does not address cross-tenant separation.

Why B is wrong: Micro-segmentation is a sound control for lateral movement within the customer's own environment, but the stated risk is another tenant on the shared host, which network segmentation between the customer's tiers does nothing to contain.

Why C is correct: In IaaS the hypervisor is the boundary that keeps co-resident tenants apart, so its hardening and isolation guarantees are what limit cross-tenant escape; in a managed public cloud the provider owns this, and the customer's concern is answered by that layer.

Why D is wrong: Encryption at rest protects stored data confidentiality, but it is proportionate to a data-theft risk, not to a co-tenancy escape, and would not stop a neighbouring workload from reaching the running virtual machine.