Network Alarm & Fault Analytics · Brief 02
Backhaul link failures · the Midwest
Upstream, not radio — and RAN alarms would have hidden it
The one line you configure
How is alarm growth splitting across element types this week — and does the pattern point at radio, at transport, or at the Norvex estate?
In a nutshell
Transport, in its own right. Backhaul link-failure alarms across the Midwest rose 55% while RAN alarms in the same region stayed flat — which is the discriminator. If radio were degrading, RAN would move first. Flat RAN with rising backhaul points upstream, and it is spreading: three of the eight affected links were clean four weeks ago.
At a glance
WHAT MOVED · Backhaul alarms against RAN alarms, by week
The flat line is the finding. Radio degradation would have moved it; a transport problem does not.
WHERE IT IS · Alarm count by link, and when it started
The three newest links are the interesting ones: whatever they share with the original five is the spread mechanism.
WHY IT MOVED · The eight affected links share one upstream segment
| Link | Region | First seen | Upstream segment |
|---|---|---|---|
| CHI-KAN-04 | Midwest | 4 wks ago | SEG-11 |
| CHI-KAN-07 | Midwest | 4 wks ago | SEG-11 |
| KAN-PHX-02 | Midwest | 3 wks ago | SEG-11 |
| CHI-STL-01 | Midwest | 1 wk ago | SEG-11 |
| CHI-STL-05 | Midwest | 1 wk ago | SEG-11 |
| 3 further links | Midwest | 2–4 wks | SEG-11 |
All eight ride SEG-11, and the three newest joined in the last week — so this is one upstream fault spreading, not eight independent link failures.
Why the weekly view matters
Backhaul alarms are intermittent, so any single day looks like noise. Rolled weekly with RAN beside it, the divergence is the finding — and the spread across links is only visible over several weeks.
So what
This is the read worth having before it becomes an outage. Nothing is isolated yet, but the spread from five links to eight in four weeks and the lengthening clear time both point the same way. Because RAN is flat, radio teams should not be pulled in — this belongs to transport, and the useful question for them is what the three newly affected links share with the original five.
The points that matter
Flat RAN is the discriminator
Backhaul up 55% with RAN at +2% in the same region rules out radio degradation as the cause.
+2%RAN alarmsIt is spreading
Eight links affected, three of which were clean four weeks ago — the pattern is propagating, not static.
8from 5 linksNothing has failed yet
No site fully isolated, so this is still preventive work rather than incident response.
0sites isolatedQuestions it already answers
What do the three newly affected links share with the original five?
Compares path, equipment generation and capacity headroom across affected and clean links to isolate the common factor.
What can this not tell us?
Not the physical cause on the path — that needs field investigation. It proves the problem is transport, not radio, and shows how fast it is spreading.