What needs fixing · Plan delivery · Capacity
From where is the problemto what to fix next.
- 01What needs fixinga board with a button per fix
- 02Plan deliveryfind the 100M router ports
- 03Capacity & call firstweeks ahead, before they call
- 04Firmware, caches, statuswhat to buy, peer and publish
to the fix
a 100M router port
per VLAN
the complaint
00See it, don't read it
The Observatory and Access Health have a new face. A top band answers in two seconds: how many problems, how many subscribers sit on affected gateways, and where on the path. Problem cards carry a mini path with the break marked and a trend line. Weather becomes a city × network heat grid with 24-hour lanes per stop.
Each problem card shows where the break is and whether it is getting worse, before you read a word.
City × network weather at a glance, with 24-hour lanes per stop of the path.
Unknown is grey, never green, and every colour comes with a glyph, so the page reads the same for colour-blind staff and on a cheap projector.
01Access Health becomes a “what needs fixing” board
Each card says where (gateway › VLAN › OLT label › line), how many subscribers, since when and the likely cause: fiber or OLT, power, gateway, 100M CPE port or Wi-Fi. And each card has a button that goes straight to the fix. Try them. illustrative demo
!All sessions dropped together
~Plans capped at 90–95 Mbps
?Outages here say “VLAN 1185”
?No link capacity entered
~Extra delay inside the home
!Line fault on one subscriber
02Plan delivery: does the line deliver what the customer pays for?
The gateway keeps each subscriber's best 10-second download window per hour, for 24 hours. Delivers: some window reached at least 85% of the plan. Capped: the same ceiling (±3%, at least 30 Mbps, below 85% of the plan) in 2 or more separate hours. Everything else is not judged: a customer who never pulls hard has no verdict, and NOC2 never calls that a bad line.
Illustrative values for one gateway. Click a bar.
03Capacity: see the next congestion weeks ahead
NOC2 stores the daily peak per VLAN. Enter each VLAN's link capacity next to its OLT label, and NOC2 shows the oversubscription (the sum of the plans sold on the VLAN divided by its link capacity) and a forecast once it has 7 days of data. Physical ports get the same check: daily peak against port speed.
Illustrative values. The dashed line is the forecast; the band is 80% of the link.
03bCall before they call
Each gateway scores the subscribers likely to need help: capped plan +3, open line fault +3, 3 or more reconnects in 24 h +2, security flag +1. Support gets a per-operator list with reason chips and a score bar, and calls them first, before the complaint.
Only the right people see names
Only the gateway's own operator sees subscriber names, and only staff allowed to see subscribers. The list is loaded on request, never stored, and every row is re-checked against the live session.
A call you make beats a call you take
"We saw your line and a technician is booked" lands better than "we are looking into it", and it keeps the queue short on a bad morning.
04Router vendors, now down to firmware families
Per CPE vendor: line delay floor (the line) against extra delay (Wi-Fi or buffering), retransmits, line faults, capped plans and the share flagged by the gateway's scanner and egress detectors. Now each vendor splits into firmware families, from the router's PPPoE handshake or DHCP fingerprint: a device-family label, never a person. Open a vendor.
05Cache offload estimate: where a cache would help
Per content owner, the share of new connections already served from a cache in or next to your network against those going over transit or IX, with the median answer time of each. Owners whose cache can't be confirmed read "not judged". Click an owner.
Video connections carry far more bytes, so a byte-based saving is usually bigger for video owners. NOC2 never turns this into Gbps or money. The anonymous benchmark "operators like you" appears only with 3 or more other operators.
06The public status page is live
status.bngsoft.com now shows its first services as "No widespread problems seen": service tiles, a big status band, a zone map that appears only during a confirmed problem (and never shows where anyone has coverage) and a 5-day history. No numbers except times, no operator names, and an invisible self-hosted bot check.
Mock-up. Services without a baseline yet still read "Not enough data".
07What operators get
A to-do list
Instead of a dashboard: each card says where, how many, since when, the likely cause, and what to press.
The right fix
The CPE port, the OLT, the VLAN link or the cache, backed by the evidence that points there.
Call first
Ring the customers most likely to complain, before they do.
Capacity, weeks ahead
Order the next link before the VLAN congests, not after the tickets.
Routers that cause tickets
Down to the firmware family, so the swap list is short and right.
Honest by default
Not judged, rolling out, estimated: each says so. Unknown is never shown as healthy.
Open your own to-do list.
We will walk through Operator Insights on a live NOC2: the board, plan delivery, the capacity forecast and the call-first list, on your questions.
Examples on this page are illustrative and anonymised. Gateway names, VLAN numbers, OLT labels, usernames, counts, times and percentages are invented demo values, not readings from any network. Vendors are shown as Vendor A–E. Google, Meta, Netflix, Akamai, Amazon and the services named are trademarks of their respective owners, named only as destinations subscribers reach; BNGSOFT is not affiliated with them.
Product page: Support & operations · Talk to us