D686 Operating Systems for Computer Scientists - Set 3 - Part 3

Correct Answer: Latency increase

Continuous verification adds authentication and authorization checks to every operation, increasing latency for system calls that previously assumed trusted internal environments inherently existed.
Option a is incorrect because while verification uses some memory, the primary overhead is the time spent performing checks not memory consumption for verification.
Option c is incorrect because storage requirements for policies are minimal compared to the runtime latency added by continuous verification of every access attempt.
Option d is incorrect because zero-trust can work locally without network traffic; the overhead is computational verification time not network communication necessarily required.

Correct Answer: Hot replication

Hot replication continuously synchronizes state between primary and standby systems, enabling instant failover without data loss when primary systems fail unexpectedly occurring.
Option a is incorrect because cold standby systems require startup and state restoration time, preventing instant failover when disasters strike primary systems suddenly.
Option c is incorrect because periodic backups have gaps between snapshots where data loss occurs, and restoration takes time preventing instant failover capability.
Option d is incorrect because manual failover requires human intervention adding delay, while hot replication enables automatic instant transition to standby systems immediately.

Correct Answer: Secure enclaves

Secure enclaves provide hardware-isolated execution environments where sensitive computations occur protected from all other software including operating system and hypervisor access.
Option a is incorrect because encrypted storage protects data at rest but doesn't isolate execution or protect data during active processing from privileged software.
Option c is incorrect because network isolation prevents external access but doesn't protect against local privileged software accessing data during computation processing phases.
Option d is incorrect because access logs record operations for auditing but don't prevent privileged software from accessing data during processing within systems.

Correct Answer: Conflict resolution

Conflict resolution algorithms must merge simultaneous edits to maintain document consistency, transforming operations to preserve all users' intentions despite concurrent modifications happening simultaneously.
Option a is incorrect because version control tracks changes over time but doesn't handle real-time simultaneous edits requiring immediate resolution for collaborative editing experiences.
Option c is incorrect because file locking prevents simultaneous editing entirely, contradicting the collaborative requirement where multiple users edit concurrently in real-time scenarios.
Option d is incorrect because backup creation preserves document states but doesn't resolve conflicts from simultaneous edits in collaborative editing environments requiring immediate consistency.

Correct Answer: Gas metering

Gas metering assigns costs to operations, limiting computation by requiring payment for resources, preventing infinite loops and resource exhaustion in blockchain environments effectively.
Option a is incorrect because simple time limits could unfairly terminate complex legitimate operations while allowing efficient malicious code to complete within limits potentially.
Option c is incorrect because memory quotas alone don't prevent computational denial-of-service attacks through intensive CPU operations within memory limits allocated to contracts.
Option d is incorrect because CPU throttling affects performance but doesn't provide the economic incentive structure that gas metering creates for efficient contract design.

Correct Answer: Heterogeneous scheduling

Heterogeneous scheduling must efficiently distribute work between different processor types with varying capabilities, optimizing for both CPU and NPU characteristics simultaneously required effectively.
Option a is incorrect because memory alignment is data layout concern not the scheduling challenge of coordinating different processor types with different capabilities fundamentally present.
Option c is incorrect because cache pollution affects performance but the primary challenge is scheduling work appropriately across heterogeneous processors not cache management alone.
Option d is incorrect because pipeline stalls occur within processors but don't represent the system-level challenge of scheduling across different processor architectures effectively needed.

Correct Answer: One-way transformation

One-way transformations create irreversible biometric templates that can verify identity but cannot recreate original biometric data, protecting privacy even if compromised completely.
Option a is incorrect because encryption is reversible with the key, potentially allowing recovery of sensitive biometric data if encryption keys are compromised later.
Option c is incorrect because access control prevents unauthorized access but doesn't protect the data itself if storage security is breached somehow occurring.
Option d is incorrect because data masking obscures portions but biometric systems need complete template data for matching, just in non-reversible format fundamentally required.

Correct Answer: Data volatility

Data volatility means memory contents disappear when power is lost, requiring special acquisition techniques to preserve evidence before system shutdown or modification occurs.
Option a is incorrect because RAM is typically smaller than disk storage; size isn't the primary challenge but rather preserving volatile contents before loss.
Option c is incorrect because RAM's faster access speed actually helps forensics; the challenge is capturing contents before they're lost or modified significantly.
Option d is incorrect because storage format differences exist but the fundamental challenge is the volatile nature requiring immediate preservation unlike persistent disk storage.

Correct Answer: Minimal attack surface

Unikernels include only code needed for specific applications, eliminating unused OS components that could contain vulnerabilities, significantly reducing potential attack surface exposure effectively.
Option a is incorrect because unikernels often have limited hardware support compared to general-purpose systems, being specialized for specific deployment environments not compatibility.
Option c is incorrect because unikernel development is more complex requiring specialized tools and knowledge compared to using general-purpose operating systems typically available.
Option d is incorrect because unikernels are highly specialized for single applications, providing less flexibility than general-purpose systems for user modifications generally.

Correct Answer: Time-partitioned indexing

Time-partitioned indexing organizes data by time ranges, optimizing recent data access while enabling efficient historical queries through partition pruning and targeted searches effectively implemented.
Option a is incorrect because hash indexing provides fast exact lookups but doesn't optimize for time-based range queries common in time-series data analysis typically needed.
Option c is incorrect because full-text indexing optimizes text searches not the numerical time-series queries and time-range operations required for sensor data analysis.
Option d is incorrect because spatial indexing optimizes geographic queries not the temporal queries that dominate time-series database workloads for sensor stream analysis.