The break is somewhere on forty kilometres of fibre
The SLA clock starts the moment the route goes dark, and a crew sent to the wrong access point costs hours that no amount of splicing speed gets back.
What actually happens
Nobody in this story is slow. The splice takes about as long as it always does. The hours go somewhere else entirely.
A route goes dark. The affected services are known within minutes, the SLA clock starts immediately, and the question that decides the next four hours is simply: where is the break.
Forty kilometres of route has a limited number of access points, and opening the wrong one is expensive in a way that compounds. The crew travels, opens, tests, finds nothing, closes, and travels again. Each wrong guess is measured in hours, not minutes.
The evidence needed to narrow it exists. OTDR traces, optical power readings at each end, the as-built route record, recent civil works notifications, and which downstream elements went dark in which order. That evidence lives in five systems owned by three teams, and assembling it takes longer than most people want to admit at two in the morning.
So the crew is often dispatched on judgement, because dispatching on judgement at least starts the travel time while the analysis continues. That is a reasonable decision under pressure and it is also why the restoration numbers look the way they do.
The second question, which route to restore first when several are affected, is usually answered by whoever shouts loudest rather than by which route carries the largest SLA exposure.
Restoration time is not splice time. It is travel plus search plus splice, and the search is the part nobody instruments.
The same night, two ways
The clock below is not about splicing faster. It is about how many access points get opened before the right one.
Illustrative, not measured. The times below model a scenario built from the patterns we see in production estates. They are not timings recorded at a named customer. The point is the shape of the clock, not the totals: check it against your own last ten incidents.
Today, dispatch then narrow
- 01:12Route goes dark. Services identified. SLA clock starts.
- 01:12 ↓ 01:48WaitingOTDR run, optical readings gathered, as-built record located. Three systems, two teams.
- 01:48Crew dispatched to the most likely access point on judgement.
- 01:48 ↓ 03:10WaitingCrew travels, opens, tests. Break is not in this span. Closes and travels on.
- 03:10 ↓ 04:05WaitingSecond access point opened. Break located.
- 05:20Splice complete. Services restored.
~4 hours · of which the splice was a small part
With Sentinel narrowing first
- 01:12Route goes dark. Sentinel opens an investigation and starts assembling location evidence immediately.
- 01:15OTDR trace, optical power at both ends, as-built route record, element-down ordering and recent civil works notifications correlated.
- 01:19Break localised to a single span between two named access points, with a confidence figure and the supporting traces attached.
- 01:20Restoration order proposed across all affected routes, ranked by SLA exposure rather than by call volume.
- 01:22Crew dispatched to the narrowed span with the trace, the as-built and the access point already on the work order.
- 03:40Splice complete. Services restored. Sherlock confirms optical levels across the route.
~2.5 hours · one access point opened, not two
The splice is not the variable. In both columns it takes about the same time, done by the same crew, with the same equipment.
The variable is how many spans get opened before the right one, and that is decided entirely by how good the location evidence was at the moment of dispatch. One wrong access point is typically the single largest line item in a restoration clock.
Sentinel assembles the location evidence in parallel rather than in sequence. OTDR, optical power, as-built records, element-down ordering and civil works notifications are queried together, so the span is narrowed while the crew is still being mobilised rather than after they have travelled.
The restoration ordering is the quieter win. When several routes are down, ranking by SLA exposure instead of by whoever escalated hardest changes which contractual penalties you actually incur, and that is a number your commercial team can price.
Why the number is what it is
The splice is not the variable. In both columns it takes about the same time, done by the same crew, with the same equipment.
The variable is how many spans get opened before the right one, and that is decided entirely by how good the location evidence was at the moment of dispatch. One wrong access point is typically the single largest line item in a restoration clock.
Sentinel assembles the location evidence in parallel rather than in sequence. OTDR, optical power, as-built records, element-down ordering and civil works notifications are queried together, so the span is narrowed while the crew is still being mobilised rather than after they have travelled.
The restoration ordering is the quieter win. When several routes are down, ranking by SLA exposure instead of by whoever escalated hardest changes which contractual penalties you actually incur, and that is a number your commercial team can price.
Sentinel assembles the location evidence in parallel rather than in sequence.
Who decides to press go
Nothing on this page dispatches a crew automatically. Sending people to a physical location is a human decision and it stays one.
Evidence assembly, span localisation, SLA exposure ranking and work order preparation all run under policy. All of it is reversible and none of it moves anyone.
The proposed dispatch, the narrowed span, the confidence figure and the supporting traces are presented to the restoration manager. A named human approves the work order before anyone travels.
The point is not to remove the judgement. It is to give the person exercising it a named span and a trace instead of forty kilometres and a hunch.
Route dark. Affected services and SLA exposure identified. Multiple downstream elements reporting loss in a specific order.
OTDR trace, optical power at both ends, as-built route record, element-down ordering and recent civil works notifications correlated into a single location hypothesis.
Break localised to one span with a confidence figure. Work order prepared with trace and access point attached. Restoration order across affected routes ranked by SLA exposure.
Civil works notifications wired into the correlation set so future breaks near known excavation are localised faster. Access point outcomes fed back into the confidence model.
This is a platform capability, not a published customer deployment for a named fibre operator. The mechanism, which is topology-aware correlation followed by governed MOP execution with pre-check, post-check, rollback and approval gating, is running in production today; see carrier-scale observability across 27,000+ devices and closed-loop network automation. The access-network scenario above applies that same mechanism to a PON estate. The timings shown are modelled, not measured at a named operator.
If the action carries no service impact
Evidence assembly, span localisation, SLA exposure ranking and work order preparation all run under policy. All of it is reversible and none of it moves anyone.
If it commits a crew or touches a live path
The proposed dispatch, the narrowed span, the confidence figure and the supporting traces are presented to the restoration manager. A named human approves the work order before anyone travels.
What Sentinel did, step by step
- ObserveRoute dark. Affected services and SLA exposure identified. Multiple downstream elements reporting loss in a specific order.
- InvestigateOTDR trace, optical power at both ends, as-built route record, element-down ordering and recent civil works notifications correlated into a single location hypothesis.
- ActBreak localised to one span with a confidence figure. Work order prepared with trace and access point attached. Restoration order across affected routes ranked by SLA exposure.
- OptimizeCivil works notifications wired into the correlation set so future breaks near known excavation are localised faster. Access point outcomes fed back into the confidence model.
Bring us your last three restorations
We will split each clock into travel, search and splice and show you where the hours sat.