Architecture & System Design Flashcards

In a microservices-based GMI measurement platform, how should service boundaries be defined?

Answer: By domain capability (e.g., calibration, measurement capture, reporting) with clear API contracts

Domain-driven boundaries align each microservice with a cohesive business capability and expose it through a versioned API contract, enabling independent deployment.

A GMI measurement system stores raw spectral data as 36-band reflectance curves. Which storage architecture best balances retrieval speed and storage efficiency for this data type?

Answer: Columnar store with compression optimized for numerical arrays

Columnar storage with compression is ideal for dense numerical arrays because it co-locates band values for fast analytical reads and compresses repetitive float data efficiently.

When designing API access for a GMI measurement platform shared across multiple print facilities, which authentication architecture provides both security and operational scalability?

Answer: Per-facility JWT tokens with role-based access control and short expiration

Per-facility JWTs with RBAC grant each facility only the permissions it needs, and short expiration limits the blast radius of credential compromise.

A GMI-certified system must synchronize color standard libraries (e.g., Pantone, HKS) across all workstations. Which distribution architecture ensures consistency without manual intervention?

Answer: Centralized library server with version-checked pull on workstation startup

A centralized library server with version checking ensures every workstation automatically pulls the latest approved standards at startup, eliminating manual drift.

In a graphic measurement pipeline, 'data provenance' means the system must record:

Answer: The origin, transformations, and lineage of every measurement value throughout the workflow

Data provenance captures the full history of a measurement — its source instrument, any applied transforms, and who acted on it — enabling full traceability.

Which load-balancing strategy is most appropriate for a GMI measurement API where some requests (e.g., full spectral job processing) take significantly longer than others (e.g., simple status queries)?

Answer: Least-connections routing so long jobs don't queue behind other long jobs on the same node

Least-connections routing sends new requests to the node with the fewest active connections, naturally avoiding overloading nodes already handling expensive spectral jobs.

A GMI measurement platform must retain raw data for 7 years for compliance while keeping storage costs manageable. Which tiered storage architecture satisfies both requirements?

Answer: Hot storage for recent 90 days, warm object storage for 1 year, cold archival storage for years 2-7

Tiering data from hot to warm to cold aligns storage cost with access frequency: recent data is fast and expensive, archival data is slow and cheap, satisfying both retention and budget requirements.