TPM Technical Expertise and Understanding 2 — Questions and Answers
Question 1: A TPM is evaluating two candidate architectures: a monolith and a microservices design. Which factor most strongly favors choosing microservices?
- Team size is small and all engineers share the same codebase
- Independent scaling of specific service components is a key business requirement (Correct answer)
- The system has low complexity and well-understood domain boundaries
- The organization lacks CI/CD maturity and deployment automation
Correct answer: Independent scaling of specific service components is a key business requirement
Microservices shine when independent scaling of discrete capabilities is needed, since each service can be deployed and scaled without affecting others.
Question 2: During a sprint, engineers discover that a critical API dependency has no SLA documentation. What should a TPM do first?
- Block the sprint until an SLA is signed
- Escalate immediately to the C-suite for resolution
- Work with the API owner to document expected availability and latency baselines (Correct answer)
- Proceed without the SLA because documenting it is the vendor's responsibility
Correct answer: Work with the API owner to document expected availability and latency baselines
Establishing documented baselines with the API owner is the immediate actionable step that unblocks the team while creating a foundation for a formal SLA.
Question 3: A TPM is reviewing a system that uses optimistic locking. In which scenario would optimistic locking cause the most problems?
- Low-traffic read-heavy reporting systems
- High-contention tables with frequent concurrent writes to the same records (Correct answer)
- Systems where most operations are append-only inserts
- Batch jobs that process records sequentially without parallelism
Correct answer: High-contention tables with frequent concurrent writes to the same records
Optimistic locking assumes conflicts are rare; under high write contention on the same rows, collision rates spike and retries degrade throughput significantly.
Question 4: Which of the following best describes 'infrastructure as code' (IaC) and its primary benefit for a TPM?
- Writing application business logic in infrastructure languages like Terraform
- Provisioning and managing infrastructure through machine-readable configuration files, enabling repeatability (Correct answer)
- Storing infrastructure diagrams in version control as documentation
- Using cloud consoles to manually replicate environments
Correct answer: Provisioning and managing infrastructure through machine-readable configuration files, enabling repeatability
IaC codifies environment definitions so they can be version-controlled, peer-reviewed, and reproducibly applied, reducing drift and manual errors.
Question 5: A TPM notices that the P99 latency of a service is 800ms while P50 is 50ms. What does this pattern most likely indicate?
- The service is consistently slow for all users
- A small percentage of requests experience significant tail latency due to outlier conditions (Correct answer)
- The median latency metric is unreliable and should be discarded
- The service is performing well because most users see 50ms responses
Correct answer: A small percentage of requests experience significant tail latency due to outlier conditions
A large gap between P50 and P99 indicates tail latency issues where a minority of requests — often hitting slow paths or resource contention — experience much worse performance.
Question 6: When should a TPM recommend using an event-driven architecture over a synchronous request-response model?
- When low latency for every individual transaction is the top priority
- When the system requires strong transactional consistency across all service calls
- When decoupling producers and consumers and absorbing traffic spikes are key requirements (Correct answer)
- When the team has no experience with message brokers or async patterns
Correct answer: When decoupling producers and consumers and absorbing traffic spikes are key requirements
Event-driven architectures decouple services through a message broker, allowing producers and consumers to scale independently and buffering load spikes.
Question 7: A TPM is assessing technical debt in a codebase. Which metric is most useful for quantifying maintenance burden?
- Lines of code per engineer
- Cyclomatic complexity and code coverage percentages (Correct answer)
- Number of open pull requests in the repository
- Average build time in the CI pipeline
Correct answer: Cyclomatic complexity and code coverage percentages
Cyclomatic complexity measures code branching paths (indicating how hard code is to test and maintain) while code coverage reveals untested areas that harbor hidden risk.
TPM Technical Expertise and Understanding 2
A TPM is evaluating two candidate architectures: a monolith and a microservices design.
Which factor most strongly favors choosing microservices?