Cloud Service Operations and Delivery
Domain 4 tests running cloud services day to day to deliver agreed outcomes and restore service quickly.
Core rules
- Incident management restores service; problem management finds and removes root causes; change enablement controls risk; service requests fulfill standard asks.
- Event management detects state changes; access management grants rights; request fulfillment uses approved catalogs and automation.
- Change types include standard, normal, and emergency. Emergency is for urgent risk, not for bypassing review.
- Cloud operations depend on provider services. Map dependencies, status feeds, and escalation paths; do not assume the provider monitors your configuration.
- Runbooks, automation, and observability should make recovery repeatable and auditable.
Worked example
Checkout latency spikes. The provider status page shows no outage, but the customer's identity provider is throttling token requests after a marketing campaign. The incident team fails over to a cached session path, restores checkout, then opens a problem record to fix capacity planning and alert thresholds. Reasoning: restore first, investigate cause second; provider health does not cover customer-side dependencies.
Traps
Closing incidents without problem follow-up; using emergency changes for convenience; assuming a provider SLA protects the customer's end-to-end service; skipping dependency mapping; and automating a broken process without guardrails.