SLO, Monitoring, and Exercise Templates — Operating Quantum Computers

This appendix provides working templates for the reliability, monitoring, and resilience-exercise practices introduced in chapters 131–133.

Quantum SLO template

Illustrative listing · yaml

service: managed-qpu-experiment
service_class: internal_research
workload_class: chemistry_vqe
measurement_window: rolling_30_days
target_success_fraction: 0.99  # illustrative; set in the service contract
eligible_event: one_evaluation_per_in_scope_job_at_its_declared_deadline
success_predicate:
  completed: true
  evidence_manifest: complete
  calibration_snapshot_max_age_hours: 24
  confidence_interval_max_width: 0.05
  cost_overrun_allowed_percent: 10
exclusions:
  - provider_announced_maintenance
  - user_cancelled_job
error_budget_policy:
  slow_burn_action: review
  fast_burn_action: freeze_noncritical_changes
  exhausted_action: pause_admission_and_open_incident
owners:
  service_owner: quantum-platform
  evidence_owner: quantum-data
  reviewer: validation-science

Diagram loads as you read

Quantum SLO template · Figure 1

View diagram source

flowchart LR
    Template[SLO template] --> Predicate[Success predicate]
    Template --> Window[Window]
    Template --> Exclusions[Exclusions]
    Template --> Budget[Error budget]
    Template --> Owners[Owners]

Error-budget review record

Illustrative listing · yaml

review_id: qeb-2026-04-001
service: managed-qpu-experiment
window_start: 2026-03-01
window_end: 2026-03-31
budget_status: watch
burn_drivers:
  - calibration_freshness_miss
  - rerun_rate_above_threshold
actions:
  - pause_low_priority_workloads_on_backend_a
  - route_chemistry_jobs_to_backend_b
  - open_problem_record_for_readout_bias
next_review: 2026-04-07

Continuous-monitoring finding

Illustrative listing · yaml

finding_id: qmon-2026-04-023
control_id: QCTRL-CAL-004
component: backend_a_scheduler
expected_state: reject_jobs_when_calibration_expired
observed_state: accepted_17_jobs_after_validity_window
severity: high
evidence:
  telemetry_query: qpu.scheduler.accepted_jobs
  evidence_objects:
    - manifest://run-abc
    - manifest://run-def
disposition: remediation_required
owner: runtime-platform
expires: 2026-04-26

Diagram loads as you read

Continuous-monitoring finding · Figure 2

View diagram source

sequenceDiagram
    participant Monitor as Monitor
    participant Finding as Finding record
    participant Owner as Owner
    participant Assessor as Assessor
    Monitor->>Finding: create with evidence
    Finding->>Owner: assign remediation
    Owner-->>Finding: update action
    Assessor->>Finding: verify closure

Exercise design packet

Illustrative listing · yaml

exercise_id: qttx-2026-02
exercise_type: tabletop
scenario: provider calibration feed stale during external benchmark
objectives:
  - classify affected workloads
  - decide whether to pause benchmark claims
  - validate evidence-completeness checks
participants:
  - runtime lead
  - validation lead
  - security lead
  - product owner
injects:
  - time: T+00
    description: calibration feed stops updating
  - time: T+20
    description: benchmark result appears unusually strong
  - time: T+45
    description: external stakeholder asks for publishable statement
stop_conditions:
  - unsafe real-system action requested
  - exercise exceeds approved duration
metrics:
  - time_to_detection
  - time_to_claim_hold
  - evidence_gap_count

Diagram loads as you read

Exercise design packet · Figure 3

View diagram source

flowchart TB
    Scenario[Scenario] --> Injects[Inject timeline]
    Injects --> Decisions[Expected decisions]
    Decisions --> Metrics[Metrics]
    Metrics --> AAR[After-action report]

After-action report skeleton

Illustrative listing · markdown

# After-Action Report

## Exercise summary

## What happened

## What worked

## What failed

## Timeline

## Evidence reviewed

## Decisions and alternatives

## Remediation items

| Item | Owner | Due date | Verification |
|---|---|---|---|

## Retest plan

Additional technical sources: [R241].