Site Reliability Engineering: SLIs, SLOs & Incident Response

Answer key & explanations

Every question in this quiz, with the correct answer marked and an explanation of why it is right. Use it to revise before or after taking the quiz above.

1. 1.What is the mathematical relationship between an SLA, an SLO, and an SLI?

   • ASLI is the contract; SLO is the measurement; SLA is the aspiration
   • BSLI is the metric measured; SLO is the internal target for that metric; SLA is the business contract with consequences for missing the target✓ Correct
   • CSLA is the monitoring dashboard; SLI is the database; SLO is the user feedback
   • DSLO and SLI are identical concepts used interchangeably by software engineers

   Correct answer: SLI is the metric measured; SLO is the internal target for that metric; SLA is the business contract with consequences for missing the target

   An SLI (Service Level Indicator) is the metric measured; an SLO (Service Level Objective) is the target reliability goal; an SLA (Service Level Agreement) is the business contract with penalties.

2. 2.What is an Error Budget in Site Reliability Engineering?

   • AThe maximum financial cost allocated to compensating customers for downtime
   • BThe acceptable threshold of unreliability (100% minus SLO) that can be spent on shipping features, changes, and deployments✓ Correct
   • CThe number of software bugs a developer is permitted to write per sprint
   • DThe memory allocation reserved for logging application stack traces

   Correct answer: The acceptable threshold of unreliability (100% minus SLO) that can be spent on shipping features, changes, and deployments

   An Error Budget is calculated as (1 - SLO); it represents the allowable level of service imperfection that teams can 'spend' on deployments, experiments, and innovation.

3. 3.What should happen when a service completely exhausts its Error Budget for a given period?

   • AAll engineers on the team must forfeit their annual bonuses
   • BFeature launches are halted, and engineering effort pivots strictly to reliability, testing, and technical debt✓ Correct
   • CThe production cluster is shut down until the next calendar month
   • DThe monitoring tool automatically lowers the SLO to avoid further alerts

   Correct answer: Feature launches are halted, and engineering effort pivots strictly to reliability, testing, and technical debt

   When an error budget is depleted, the agreed-upon policy mandates halting risky feature launches to refocus engineering bandwidth on stabilizing the platform.

4. 4.What is a major advantage of Multi-Window, Multi-Burn-Rate alerting compared to simple threshold alerting?

   • AIt alerts engineers only during regular office working hours
   • BIt alerts based on the rate at which the error budget is being consumed over multiple time windows, avoiding false positives from transient blips✓ Correct
   • CIt automatically repairs corrupted server configurations
   • DIt requires no monitoring instrumentation in application code

   Correct answer: It alerts based on the rate at which the error budget is being consumed over multiple time windows, avoiding false positives from transient blips

   Burn rate alerts track the speed of error budget consumption across short and long time windows simultaneously, catching critical outages fast while eliminating noisy transient alerts.

5. 5.What is Toil in SRE methodology?

   • AAny work that involves writing production software code
   • BRepetitive, manual, automatable work that scales linearly with service growth and lacks enduring engineering value✓ Correct
   • CThe time spent participating in interview loops for hiring candidates
   • DThe physical energy consumed by server hardware in a data center

   Correct answer: Repetitive, manual, automatable work that scales linearly with service growth and lacks enduring engineering value

   Toil is manual, repetitive, administrative, and automatable operational work that scales linearly with service size; SRE aims to cap toil at under 50% of an engineer's time.

6. 6.What is the primary objective of a Blameless Postmortem?

   • ATo identify and reprimand the engineer who caused the outage
   • BTo uncover systemic and technical root causes without assigning individual fault, implementing preventative safeguards✓ Correct
   • CTo produce a legal document used to defend against customer lawsuits
   • DTo satisfy external regulatory compliance auditors

   Correct answer: To uncover systemic and technical root causes without assigning individual fault, implementing preventative safeguards

   Blameless postmortems assume well-intentioned staff make mistakes within flawed systems; they focus on systemic causes to prevent recurrence without punishing individuals.

7. 7.Why is Mean Time to Recovery (MTTR) generally considered a more actionable metric to optimize than Mean Time Between Failures (MTBF) in complex distributed systems?

   • AComplex distributed systems have too many moving parts to prevent all failures; fast detection and recovery minimize user impact more effectively✓ Correct
   • BMTBF cannot be calculated using standard arithmetic
   • CMTTR applies only to hardware infrastructure, whereas MTBF applies only to software
   • DOptimizing MTTR requires no monitoring systems or on-call staff

   Correct answer: Complex distributed systems have too many moving parts to prevent all failures; fast detection and recovery minimize user impact more effectively

   In large distributed systems, component failures are an inevitability; investing in rapid mitigation, observability, and rollback (MTTR) preserves user experience better than trying to achieve zero failures.

8. 8.What does a Circuit Breaker pattern prevent during an ongoing incident involving an overloaded downstream service?

   • AIt prevents database queries from using indexes
   • BIt stops upstream callers from continuing to hammer the struggling service, preventing cascading failure across the architecture✓ Correct
   • CIt restarts the host operating system kernel automatically
   • DIt prevents unauthorized users from logging into the portal

   Correct answer: It stops upstream callers from continuing to hammer the struggling service, preventing cascading failure across the architecture

   A circuit breaker trips when failure rates rise, immediately returning errors or fallbacks to callers without hitting the downstream service, giving it space to recover.

9. 9.What role does the Incident Commander (IC) fulfill during a major production incident?

   • AWriting the actual code fix while on the war room call
   • BHolding ultimate operational authority, directing response logistics, delegating tasks, and maintaining focus across the team✓ Correct
   • CCommunicating directly with external news media and shareholders
   • DExecuting manual database backup scripts

   Correct answer: Holding ultimate operational authority, directing response logistics, delegating tasks, and maintaining focus across the team

   The Incident Commander leads the incident response, delegating roles (investigators, communications), coordinating decisions, and ensuring the team stays organized without getting hands-on in the code.

10. 10.What is Alert Fatigue, and what is its primary danger?

    • APhysical exhaustion caused by staring at bright monitoring screens
    • BEngineers becoming desensitized to frequent, non-actionable notifications, leading to missed or ignored critical alerts✓ Correct
    • CMonitoring systems crashing due to high telemetry volumes
    • DPager systems running out of cellular transmission credits

    Correct answer: Engineers becoming desensitized to frequent, non-actionable notifications, leading to missed or ignored critical alerts

    Alert fatigue happens when engineers are bombarded with non-actionable or false alerts; they become desensitized and may miss or delay responding to genuine outages.

11. 11.In the context of the Four Golden Signals (Latency, Traffic, Errors, Saturation), what does Saturation measure?

    • AThe volume of network requests hitting the load balancer
    • BHow full a service or resource is, tracking constrained components like memory, CPU, or pool capacity✓ Correct
    • CThe percentage of HTTP requests returning 500 error codes
    • DThe number of hours an on-call engineer has worked

    Correct answer: How full a service or resource is, tracking constrained components like memory, CPU, or pool capacity

    Saturation measures system fullness—the fraction of constrained resources in use (e.g., CPU, memory, database pool usage, disk I/O queue depths).

12. 12.Why is it best practice to measure latency using percentiles (e.g., p95, p99) rather than simple arithmetic averages (mean)?

    • AAverages cannot be calculated over large streaming datasets
    • BAverages obscure tail latency outliers, masking significant performance degradation experienced by substantial user subsets✓ Correct
    • CPercentiles are strictly required by ANSI SQL standards
    • DModern monitoring tools cannot compute averages

    Correct answer: Averages obscure tail latency outliers, masking significant performance degradation experienced by substantial user subsets

    Averages hide extreme values; an average latency of 100ms could conceal the fact that the 99th percentile of users is waiting five seconds for their requests.

13. 13.What is exponential backoff with jitter in distributed retry strategies?

    • AA method for compressing log lines using dynamic bit-shifting
    • BProgressively doubling retry delays while adding random variance to prevent retrying clients from clumping into synchronized waves✓ Correct
    • CA protocol for restarting servers concurrently in parallel
    • DRetrying a failed HTTP request immediately without any delay

    Correct answer: Progressively doubling retry delays while adding random variance to prevent retrying clients from clumping into synchronized waves

    Exponential backoff doubles wait times between successive retries, while jitter introduces random noise to prevent retrying clients from hammering the server in synchronized lockstep.

14. 14.What is the purpose of an operational runbook (or playbook)?

    • AA collection of marketing documents outlining future software releases
    • BStep-by-step procedures and diagnostics designed to help on-call engineers triage, troubleshoot, and mitigate specific system alerts✓ Correct
    • CAn employment contract detailing on-call compensation terms
    • DAn automated script that replaces human software engineers entirely

    Correct answer: Step-by-step procedures and diagnostics designed to help on-call engineers triage, troubleshoot, and mitigate specific system alerts

    A runbook gives on-call engineers clear, actionable instructions, diagnostic steps, and mitigation commands to resolve specific alerts quickly under stressful conditions.

15. 15.What is Chaos GameDay in an organization?

    • AAn esports tournament organized for company employees
    • BA structured exercise where teams intentionally trigger controlled production failures to validate system resilience and team response readiness✓ Correct
    • CA day when developers are permitted to deploy unreviewed code directly to production
    • DA scheduled 24-hour maintenance window where all services are offline

    Correct answer: A structured exercise where teams intentionally trigger controlled production failures to validate system resilience and team response readiness

    A GameDay is a scheduled, coordinated drill where engineers simulate real-world failure modes to observe system behavior and validate incident response procedures.

16. 16.What does Graceful Degradation mean in resilient systems design?

    • AA server shutting down immediately whenever an error is detected
    • BThe ability of a system to maintain core operations by disabling non-critical features during outages or resource exhaustion✓ Correct
    • CThe natural deterioration of hardware components over several years of operation
    • DSlowly reducing network bandwidth to all users equally

    Correct answer: The ability of a system to maintain core operations by disabling non-critical features during outages or resource exhaustion

    Graceful degradation ensures that when components fail, the system sheds non-essential features (e.g., hiding personalized recommendations) while keeping core functionality working.

More free quizzes

• ITDistributed Systems: Consensus & Fault ToleranceDeep dive into distributed systems engineering, covering consensus protocols (Raft/Paxos), the CAP theorem, and vector clocks.16 questions
• ITLLM Engineering: Fine-Tuning, Alignment & ServingTest your knowledge of transformer architectures, PEFT techniques (LoRA), RLHF, vLLM, and production deployment strategies.16 questions
• ITApplied AI: Systems, Embeddings & RAGAssess your understanding of practical machine learning architectures, embedding spaces, and Retrieval-Augmented Generation systems.16 questions