Two Time Zones Do Not Establish Incident Coverage

Calculate dated response windows, overlap and uncovered intervals, then define acknowledged incident handover. Includes a reproducible fictional UTC coverage example.

Calculate incident coverage from accepted response windows and qualified responders, not team locations. Convert a dated schedule to one time basis, take the union of its usable intervals and show the gaps. Then require the incoming responder to acknowledge each open incident. Two teams working in different time zones can still leave most of a day without an agreed response.

This article helps a buyer and service owner evaluate a distributed AWS operating arrangement. It does not describe Ampity staffing, promise continuous coverage or recommend a response time. The UTC windows, people and incident below are fictional teaching inputs. No on-call system, contact or AWS account was exercised.

Define what covered means for this workload

A staffed window is not necessarily an incident-response window. A person can be available for project work without holding the skills, access or authority needed for an outage. Define the covered service, incident classes, accepted channel, acknowledgement expectation and permitted initial actions. Record the responsible team and backup route for that boundary.

Keep acknowledgement, investigation and restoration separate. A responder can acknowledge a case while waiting for an authorized customer decision. A service restoration target needs a workload-specific recovery design and evidence. It should not appear as an automatic consequence of overlapping working hours.

AWS operational-readiness guidance includes personnel, procedures and support planning in readiness. For this review, check the actual responder roster, available tools and workload knowledge alongside the calendar. A scheduled but unqualified primary leaves a capability gap even if a coverage spreadsheet colors the hour green.

Define exceptions before calculating: weekends, holidays, leave, training, travel and degraded communications. A regular weekday schedule cannot establish annual coverage. A backup name without an accepted duty is an unresolved dependency. Keep the buyer's own retained response responsibilities visible rather than assuming the external team absorbs every gap.

Put the actual date and time basis in the record

Use the local timezone identifier, effective date and UTC conversion for each window. A fixed offset copied from one month may be wrong for another date where daylight-saving rules apply. Confirm the conversion with the scheduling system used for the real service, and retain its dated output. The example avoids conversion claims by supplying UTC directly.

Use half-open intervals: a window includes its start but ends just before its end timestamp. Thus one interval ending at 12:00 and the next starting at 12:00 can form continuous scheduled coverage without a positive overlap period. A handover needing discussion time requires actual overlap or another accepted process. A shared endpoint alone does not create that time.

For windows crossing midnight, split the calculation at the UTC day boundary or retain a complete multi-day timeline. Do not compare only clock labels and discard dates. Repeating a daily pattern also needs a policy for weekends and holidays; an eight-hour weekday interval is not eight hours on every calendar day.

Reproduce one day's coverage calculation

For fictional day D1, Team A accepts response duty from 04:00 to 12:00 UTC and Team B from 10:00 to 18:00 UTC. Assume one responder from each team, both qualified for the same workload, severity and initial actions. There are no breaks or exceptions in this teaching case. Each window has eight hours, so the sum is 16 responder-hours.

The overlap is 10:00 to 12:00, or two hours. The covered union is 04:00 to 18:00, or 14 clock-hours: 8 + 8 - 2 = 14. The uncovered intervals are 00:00 to 04:00 and 18:00 to 24:00, totaling ten hours: 4 + 6 = 10. Two available responders during overlap do not fill an hour outside either window.

If Team B instead starts at 12:30 and ends at 20:30, there is no overlap. The union becomes 16 hours, but a 30-minute gap appears from 12:00 to 12:30. Its other gaps are four hours before Team A and three-and-a-half hours after Team B. Total uncovered time is still eight hours. More total coverage has not created a usable live handover.

If Team B cannot access the affected environment, its interval cannot count as equivalent action coverage under the original definition. It might still provide communication or triage. Calculate those services separately rather than claiming 14 hours of recovery capability from a contact roster.

Transfer an incident only after the receiver accepts it

Assume incident I6 begins at 11:50 UTC, ten minutes before Team A's window ends. Team A records the affected workflow, observed impact, completed actions, unresolved effects, next permitted action and current decision owner. Team B must acknowledge the incident and those outstanding duties through the agreed channel. Sending a message at 11:59 is not the same as receiving acknowledgement.

The handover should identify evidence by controlled reference, not paste credentials or customer payloads into a broad channel. Include the last known service state, uncertain external actions and the next observation. If a payment or export timed out, the incoming responder must know whether replay could duplicate an effect. “Restart pending” without that context can create a new incident.

AWS escalation guidance calls for defined ownership, escalation conditions and third-party coordination. Use those principles to specify what happens when Team B does not acknowledge. The outgoing responder follows the accepted escalation or fallback arrangement. This article cannot impose indefinite overtime or grant emergency production authority.

If the fallback cannot provide qualified response, record a coverage failure and invoke the agreed customer communication process. An honest gap allows a service owner to select containment or a different service arrangement. A ticket silently assigned to an unavailable person conceals the same operational risk without resolving it.

Choose the arrangement that fits the actual gaps

An explicitly limited-hours service can be suitable for a workload that tolerates delayed response and has an accepted outside-window policy. State when requests wait and which customer duties continue. Do not describe that as continuous monitoring with guaranteed action unless the actual arrangement establishes it.

A separate on-call duty can cover selected gaps, but it needs accepted responders, compensation or supplier terms, tooling, access, escalation and exercises. A phone number alone is not an operating model. A narrower emergency-only arrangement can also be valid if ordinary support remains limited and the distinction is clear to users.

Automation can detect or contain known conditions outside staffed windows. It introduces its own permissions, failure modes, observation and recovery needs. An alarm firing does not establish human response; an automatic action does not establish business recovery. Keep automated containment and qualified human coverage as different rows in the service definition.

Next action: Recheck the schedule after a real change

Repeat the dated calculation when a roster, timezone rule, holiday calendar, duty scope or access arrangement changes. Verify the alert channel and handover using an approved non-disruptive exercise. Include a missed acknowledgement and an unavailable primary so that the fallback is more than a name. Record what the exercise actually established, not a universal future response guarantee.

The cloud reliability playbook owns workload reliability and recovery evidence. The enterprise incident-authority article owns who may perform each action across customer, vendor and partner. This schedule review answers a narrower question: when a qualified owner is actually committed to receive the work, and how responsibility crosses the boundary.

Bring the next seven dated service days, one actual duty definition and one fictional open incident to the operating review. Calculate the gaps first. Then ask the incoming responder to demonstrate acknowledgement and locate the next permitted action. Resolve that specific gap before expanding the support promise or using geography as a substitute for coverage.