Cloud Engineer Research & Evidence-Based Practice 5 — Questions and Answers
Question 1: A cloud team is evaluating whether multi-cloud strategy reduces vendor lock-in risk. Which type of evidence would most strongly support or refute this hypothesis?
- Opinions from cloud engineers on LinkedIn
- Longitudinal case studies documenting actual migration timelines, costs, and operational complexity from organizations that pursued multi-cloud (Correct answer)
- Marketing materials from multi-cloud management vendors
- Theoretical architecture diagrams showing portability patterns
Correct answer: Longitudinal case studies documenting actual migration timelines, costs, and operational complexity from organizations that pursued multi-cloud
Longitudinal case studies from organizations that actually executed multi-cloud strategies provide real-world evidence of costs, benefits, and trade-offs over time.
Question 2: When designing a load test to generate evidence for a cloud capacity planning decision, which factor is most critical to control?
- The geographic location of the load testing servers
- Ensuring the load profile (request types, concurrency, think time) accurately reflects production traffic patterns (Correct answer)
- Using the same cloud provider for both load generators and the system under test
- Running the test on the first Monday of each month for consistency
Correct answer: Ensuring the load profile (request types, concurrency, think time) accurately reflects production traffic patterns
A load test only yields valid capacity planning data if the synthetic traffic accurately mimics real-world traffic distribution and patterns.
Question 3: Which practice best embodies the 'measure twice, cut once' principle in cloud infrastructure changes?
- Running every change through two separate code reviewers
- Collecting baseline metrics before any change and defining rollback criteria, then deploying and comparing against baseline (Correct answer)
- Deploying to two regions simultaneously to compare outcomes
- Testing the change in staging twice before promoting to production
Correct answer: Collecting baseline metrics before any change and defining rollback criteria, then deploying and comparing against baseline
Establishing a pre-change baseline and clear rollback criteria creates the evidence needed to evaluate whether the change achieved its goal without harm.
Question 4: A cloud engineer wants to apply findings from Google's Site Reliability Engineering (SRE) book to their small startup. What critical thinking step is most important?
- Implement all SRE practices immediately as Google represents best practice
- Assess which SRE principles apply at your scale and organizational maturity, adapting rather than directly copying Google's context (Correct answer)
- Discard SRE guidance because Google's infrastructure is incomparable to a startup
- Hire former Google SREs before implementing any SRE practices
Correct answer: Assess which SRE principles apply at your scale and organizational maturity, adapting rather than directly copying Google's context
Practices from hyperscalers often embed assumptions about scale, team size, and resources that don't transfer directly to smaller organizations, requiring contextual adaptation.
Question 5: What is 'Hawthorne effect' and why is it relevant to cloud performance experiments?
- A bias where systems perform better during observation periods due to increased attention, potentially inflating experiment results (Correct answer)
- The tendency for cloud costs to decrease when engineers are monitoring dashboards more closely
- Performance degradation caused by monitoring agent overhead during tests
- The improvement in reliability that occurs when a new team takes ownership of a service
Correct answer: A bias where systems perform better during observation periods due to increased attention, potentially inflating experiment results
The Hawthorne effect can cause systems or teams to perform atypically during a monitored experiment period, making results less representative of normal operation.
Question 6: A cloud team wants to build an evidence-based culture around incident response. Which leading indicator is best supported by research on high-reliability organizations?
- Number of incidents escalated to senior engineers per quarter
- Rate of near-miss reporting and action item closure from post-mortems (Correct answer)
- Average tenure of on-call engineers
- Number of runbooks documented in the wiki
Correct answer: Rate of near-miss reporting and action item closure from post-mortems
High-reliability organization research shows that near-miss reporting rates and post-mortem action item closure are the leading indicators most predictive of improving safety and resilience.
Question 7: When publishing an internal research report on a cloud cost optimization experiment, which element most strengthens the report's credibility and reusability?
- Executive summary endorsed by senior leadership
- A methodology section detailing data sources, collection period, tools used, exclusions, and assumptions (Correct answer)
- Colorful charts and visualizations for executive presentations
- Comparison to a competitor's publicly reported cloud spend
Correct answer: A methodology section detailing data sources, collection period, tools used, exclusions, and assumptions
A detailed methodology section allows others to critique the approach, reproduce the analysis, and understand the conditions under which the findings apply.
Cloud Engineer Research & Evidence-Based Practice 5
A cloud team is evaluating whether multi-cloud strategy reduces vendor lock-in risk.
Which type of evidence would most strongly support or refute this hypothesis?