Architect Certification Mock Exam Day 71: 75 Questions to Test Knowledge Retrieval

Overview

This article presents a comprehensive closed-book diagnostic mock exam (Day 71 of the "Score Architect" series) designed for software architect certification preparation. The test contains 75 single-choice questions worth 1 point each, with a 120-minute time budget. It is an original diagnostic simulation, not drawn from past exam papers or official prediction materials. The questions sample previously studied content and include targeted diagnostic items for concepts that learners have repeatedly asked about. The distribution does not reflect actual exam weightings and does not claim full textbook coverage.

Test Structure and Topics

Computer Architecture and Organization (Questions 1–3, 15)

CPU execution time calculation given instruction count, CPI, and clock frequency.

Bus theoretical bandwidth from clock frequency, bus cycles per transfer, and data width.

Maximum addressable space for a 32-bit byte-addressed address bus.

Classic RISC characteristics: load/store architecture, register-to-register operations.

Memory Hierarchy and Storage (Questions 4–9)

Spatial vs. temporal locality demonstrated by sequential array access and repeated variable access.

Traditional DMA operation: CPU sets parameters, DMA controller performs bulk transfer, interrupts on completion.

RAID 10 fault tolerance: which disk pair can fail without data loss (mirrored pairs).

Paging: logical address split into page number and offset for 64 KiB space, 1 KiB pages.

Page table translation: logical address 2010H with given page-to-frame mappings.

FIFO page replacement with 2 frames for reference string 0,0,1,1,3,1,2.

Operating Systems and Concurrency (Questions 10, 23)

Deadlock prevention by enforcing lock acquisition order breaks the circular-wait condition.

Optimistic locking with version check: zero affected rows means no row satisfied all conditions; cause must be investigated.

Networking (Questions 11–14)

IPv4 subnet calculation for 192.168.10.130/26: network address and usable host count.

First packet to remote subnet via default gateway with no ARP cache: ARP for gateway MAC, frame to gateway MAC, IP destination unchanged.

TCP flow control vs. congestion control: flow control protects receiver buffer, congestion control protects network.

Recursive DNS resolver with valid cached record: answers directly from cache without querying root servers.

Database Theory and Design (Questions 16–19)

Relation R(StudentID, Name, CourseID, Grade) with FDs StudentID→Name, (StudentID,CourseID)→Grade: partial dependency of Name on candidate key violates 2NF.

3NF condition for non-trivial FD X→A: X is a superkey or A is a prime attribute.

Relation R(Student, Course, Teacher) with FDs Teacher→Course, (Student,Course)→Teacher: satisfies 3NF but not BCNF because Teacher is not a superkey while Course is prime.

Lossless decomposition of R(A,B,C) with A→B into R1(A,B) and R2(A,C): common attribute A determines all attributes of R1.

Database Indexing, Transactions, and Scaling (Questions 20–27)

InnoDB secondary index on name covering SELECT id,name: secondary index leaf stores name and primary key id, enabling index-only scan.

Composite index (patient_id, created_at) for equality on patient_id and range on created_at: locate patient's index range then filter by time.

Keyset pagination for DESC sort: filter with created_at < t OR (created_at = t AND id < k).

Read routing with async replication: routine reports read replicas; post-commit confirmation reads primary or consistent read path.

Wide-column model: row key locates row, column family groups columns, dynamic column qualifiers appear as needed.

Star to snowflake schema: fact table joins dimension tables; normalized dimensions (e.g., store→city→region) add hierarchy levels.

Lambda architecture merge: new batch result replaces covered period; only uncovered real-time increments are added.

Requirements Engineering and Modeling (Questions 28–31)

Unclear UI requirements: build prototype, gather feedback, iteratively clarify.

Baseline change request: record change, analyze scope/cost/schedule/risk, approve, update baseline and traceability.

UML composition for Order–OrderItem: solid diamond on Order (whole) connected to OrderItem (part) with solid line.

Time-ordered service calls: sequence diagram.

Design Principles and Patterns (Questions 32–33)

Dependency Inversion: core logic depends on payment interface; vendor-specific adapters implement interface.

Runtime-selectable discount algorithms per patient type: Strategy pattern.

Testing and Quality Assurance (Questions 34–40)

Boundary value analysis for integer 18–100: 17,18,19,99,100,101.

V-model: high-level design corresponds to integration testing.

Single if(A && B) with (T,T) and (F,F) covers both decision outcomes but not MC/DC (independent condition influence).

CMMI: defined standard processes tailored per project, no statistical control → Level 3 Defined.

OS/database adaptation without functional change: adaptive maintenance.

Executable test case minimum: test inputs/conditions and expected results.

Quality attribute scenario: 95% of requests within 2 seconds is the response measure.

Architecture Concepts and Evaluation (Questions 41–47)

GovCloud deployment (constraint), P95≤2s (quality goal), caching/multi-instance (architectural decisions).

Log pipeline parse→filter→transform→aggregate: pipe-filter style.

Multiple knowledge sources collaborating on shared blackboard: blackboard style.

4+1 views: deployment nodes, network, software-to-node mapping → physical view.

ATAM: thread pool size affecting latency = sensitivity point; synchronous replication affecting consistency/latency/availability = tradeoff point.

Utility tree leaf scoring: focus limited evaluation effort on important and challenging scenarios.

Economic comparison of architectural improvements: CBAM (Cost Benefit Analysis Method).

Enterprise Architecture and Integration (Questions 48–50)

TOGAF ADM phases B, C, D: Business Architecture; Information Systems Architectures (Data & Application); Technology Architecture.

SOA core: organize capabilities as services with contracts, explicit interfaces for loose coupling.

Bounded contexts: define model/term applicability boundaries so each model stays consistent within its context.

Cloud Native and Infrastructure (Questions 51–53, 61–63)

Pod running but not ready for traffic, no restart: readiness probe.

Hard real-time 20 ms deadline on RTOS: analyze worst-case execution time, scheduling, blocking, interrupt latency to verify deadline feasibility.

Hadoop (batch ecosystem) vs. Lambda (batch+speed serving architecture): complementary, can be combined.

Retry storm: 1000 req/s with up to 5 immediate retries → ~6000 attempts/s under sustained failure.

Sidecar service mesh path: Service A → local proxy → remote proxy → Service B; control plane pushes config to proxies.

Reliability, Capacity, and Project Management (Questions 63–70)

Availability = MTBF/(MTBF+MTTR) = 999/(999+1) = 99.9%.

Parallel system with two 0.9-reliability components, perfect switching: 1-(1-0.9)^2 = 0.99.

16 instances (8 per AZ) at 420 QPS each; one AZ fails → 8 instances ≈ 3360 QPS, cannot claim full 6000 QPS peak capacity.

PERT estimate: (optimistic + 4×most likely + pessimistic)/6 = (2+20+8)/6 = 5 days.

Critical path: A(3)→C(5) = 8 days; A(3)→D(2) = 5 days; B(4)→D(2) = 6 days. Project ends after C and D → 8 days.

Earned value: PV=100, EV=80, AC=90 → SPI=0.8 (behind), CPI≈0.889 (over budget).

EAC = BAC/CPI = 200/0.8 = 250.

End-to-end procurement process with cross-department responsibility and bottleneck monitoring: Business Process Management (BPM).

Professional English Terminology (Questions 71–75)

Idempotent: repeating with same request identifier creates no additional business effects.

Tradeoff point: architectural decision affecting multiple quality attributes requiring interaction analysis.

Tail latency: e.g., P95 for fixed workload performance testing.

Recovery time objective (RTO): system shall recover within five minutes after AZ failure.

Traceable: requirements must remain traceable to link design decisions and tests to business needs.