TPM Technical Expertise and Understanding 3 — Questions and Answers
Question 1: A TPM's team is building a system that must handle 10,000 requests per second at peak. Which capacity planning approach is most appropriate?
- Provision for average load and add capacity reactively when alerts fire
- Model peak load with a safety margin, use load testing to validate, and define auto-scaling policies (Correct answer)
- Overprovision by 10x to eliminate any risk of overload
- Rely entirely on the cloud provider's default autoscaling without load testing
Correct answer: Model peak load with a safety margin, use load testing to validate, and define auto-scaling policies
Validated load testing against modeled peak demand, combined with defined auto-scaling policies, gives evidence-based confidence without unnecessary over-provisioning.
Question 2: What is the purpose of a circuit breaker pattern in distributed systems?
- To encrypt traffic between microservices to prevent data leaks
- To stop cascading failures by halting calls to a degraded downstream service after a failure threshold (Correct answer)
- To load-balance traffic evenly across all service instances
- To cache responses from upstream services to reduce database load
Correct answer: To stop cascading failures by halting calls to a degraded downstream service after a failure threshold
A circuit breaker opens when error rates exceed a threshold, failing fast instead of waiting for timeouts, which prevents one failing service from taking down dependents.
Question 3: A TPM is choosing between horizontal and vertical scaling for a stateful database tier. Which statement is most accurate?
- Horizontal scaling is always easier for stateful databases than vertical scaling
- Vertical scaling (larger instance) is often simpler for stateful databases, while horizontal scaling requires sharding strategies (Correct answer)
- Stateful databases scale identically to stateless application tiers
- Horizontal scaling is preferred for databases because it requires no schema changes
Correct answer: Vertical scaling (larger instance) is often simpler for stateful databases, while horizontal scaling requires sharding strategies
Stateful databases carry data locality and consistency constraints that make horizontal scaling complex, often requiring sharding or replication strategies that vertical scaling avoids.
Question 4: In a CI/CD pipeline, what distinguishes continuous delivery from continuous deployment?
- Continuous delivery automates testing only; continuous deployment also automates building
- Continuous delivery requires a manual approval gate before production; continuous deployment releases to production automatically (Correct answer)
- Continuous deployment is only applicable to cloud-native applications
- Continuous delivery deploys to production on a fixed weekly schedule
Correct answer: Continuous delivery requires a manual approval gate before production; continuous deployment releases to production automatically
Continuous delivery keeps software always releasable and uses a manual gate before production, while continuous deployment removes that gate and automatically ships every passing build.
Question 5: A TPM must communicate the risk of not addressing a known SQL injection vulnerability to a non-technical executive. What is the most effective framing?
- Explain the CVE score and OWASP ranking in detail
- Describe the vulnerability in code-level terms and share the exploit proof-of-concept
- Quantify potential business impact: data breach costs, regulatory fines, and reputational damage (Correct answer)
- Defer the discussion until security engineers can present a technical deep-dive
Correct answer: Quantify potential business impact: data breach costs, regulatory fines, and reputational damage
Executives respond to business impact framing — translating technical risk into financial and reputational consequences drives prioritization decisions.
Question 6: Which data consistency model does a TPM need to understand when working with distributed NoSQL databases that prioritize availability?
- ACID transactions with two-phase commit
- Eventual consistency, where replicas converge over time without guaranteeing immediate synchrony (Correct answer)
- Linearizability, ensuring every read reflects the most recent write globally
- Strict serializability enforced by a single global lock
Correct answer: Eventual consistency, where replicas converge over time without guaranteeing immediate synchrony
Many NoSQL systems designed for high availability (per the CAP theorem) trade strong consistency for eventual consistency, meaning replicas may temporarily diverge.
Question 7: A TPM is reviewing a proposed caching strategy. Which scenario is LEAST suitable for aggressive caching?
- Static product catalog data that changes weekly
- User-specific account balances that must reflect real-time transactions (Correct answer)
- Publicly available reference data like country code lists
- Pre-computed aggregated analytics reports refreshed nightly
Correct answer: User-specific account balances that must reflect real-time transactions
Account balances require real-time accuracy for financial correctness; stale cached values could lead to overdrafts, errors, or fraud.
TPM Technical Expertise and Understanding 3
A TPM's team is building a system that must handle 10,000 requests per second at peak.
Which capacity planning approach is most appropriate?