Managed monitoring for wireless ISPs

Your subscribers shouldn't be your alerting system.

We poll every radio, router, switch and rectifier across your towers around the clock, and tell you what's degrading while it's still fixable. No new hardware. No platform for your team to learn.

Price it and start See what we watch 14-day pilot · no card · cancel anytime after
Incident replay · 60× 02:36:00
Detection timeline
02:38
Signal drifts outside 14-day baseline
02:44
Retransmits climbing, second poll confirms
02:47
Engineer triages, on-call notified with cause
03:11
First subscriber call — didn't need to happen

Replay of a typical backhaul degradation, sped up 60×. The gap between the alert and the first phone call is the product.

What we watch

Built for what's actually on your towers.

Most monitoring tools were written for data centres and then pointed at a WISP. Ours started on wireless plant — signal levels, modulation drops, PoE budgets, float voltage, battery runtime. The things that actually take your network down.

RF and backhaul

Every link measured against its own 14-day baseline, so a slow slide across three weeks is as visible as a hard failure at 2am.

  • signal / SNR per chain
  • MCS and modulation state
  • retransmit and drop rate
  • airtime and capacity headroom
  • registered subscriber count

Power and site health

The overnight, mid-winter failures. Where most WISP monitoring is completely blind and the cost of being wrong is a dead site.

  • rectifier output and alarms
  • battery voltage and runtime
  • PoE budget headroom
  • enclosure temperature
  • mains failure and generator state

Routing and edge

Core and distribution, with thresholds set against your topology and traffic pattern rather than a vendor's defaults.

  • CPU, memory, disk
  • interface errors and discards
  • BGP and OSPF session state
  • queue drops and CCQ
  • uptime and config change events
Cambium ePMPUbiquiti airMAX LTUUniFiMimosa MikroTik RouterOSAviat Eltek & Delta rectifiers anything with an SNMP OID

Coverage

Your overnight is our business day.

The hardest shift to staff in a US network operation is the one nobody wants. It isn't a hard shift for us — we work it in daylight, awake, on a normal schedule. That's the entire reason this arrangement works, and it's why we can price it the way we do.

Staffed by Nightpoll Your own team on shift
21:00 → 09:00 ET

Twelve hours where a failed backhaul otherwise sits unnoticed until the morning crew finds it — or a subscriber does first.

03:00 → 15:00 SAST

The same twelve hours, mid-morning to mid-afternoon for us. Eyes on the board, not a pager next to a bed. Full 24/7 cover available as an add-on.

Pricing & sign-up

Build your plan. Start today.

A monthly platform fee plus a per-device rate that steps down as you grow. One-time onboarding covers template build, threshold tuning and 14 days of baselining — waived on annual plans. No minimum term. Card through Stripe, or request an invoice for AP.

1 · Choose your level
2 · How many devices?
Minimum 15 devices. Rate steps down past 100 and 300.
3 · Add-ons
4 · Billing

Prefer to own it? Build & hand over

We deploy Zabbix on your own servers, build templates for your full device fleet, generate site and interface documentation from your configs, and train your team to run it. You own all of it — no licence, no lock-in, no monthly fee.

$5,900
one-time
Buy the build
STEP 01

Pay and share access

Checkout takes about a minute. You'll get an access request listing exactly what we need — read-only SNMP v3, a restricted API user, and a VPN endpoint you control.

STEP 02

We discover and baseline

We sweep your ranges, map the fleet, and run 14 days of baselining so thresholds fit your network instead of a vendor default. You watch it fill in.

STEP 03

Alerting goes live

Alerts route to Slack, Teams, SMS or your ticket queue. Your named engineer walks you through the board, and the overnight watch starts that night.

Who runs it

A network engineer, not a call centre.

Nightpoll is run by engineers who operate a live wireless network every day — roughly twenty tower sites of Cambium, Ubiquiti, Mimosa and MikroTik plant, with the rectifiers, batteries and subscriber complaints that come with it.

That matters more than it sounds. We wrote our own SNMP template library because the off-the-shelf ones alert on the wrong things and stay silent on the right ones. We know a Force 425 reports differently to a Force 300, that a UniFi PoE budget alert fires before anything is actually wrong, and that a rectifier float voltage drift at 03:00 is worth waking someone for.

When you send us a device list, the person reading it has configured that hardware. There is no tier one.

Operating experience
Live WISP network, ~20 tower sites, daily NOC and field escalation
Monitoring stack
Zabbix, purpose-built SNMP templates per device class
Vendor depth
Cambium, Ubiquiti, Mimosa, MikroTik, Aviat, Eltek, Delta
Escalation
Named engineer, direct line, no ticket queue in between

Questions

The ones operators ask first.

What exactly am I paying for when I check out?

A monthly subscription at the price shown, starting after your 14-day pilot. The pilot covers up to 25 devices at no charge, and your card isn't billed until you confirm you want to continue. After that it renews monthly and you can cancel with 30 days' notice from your account.

Do we have to move off what we're already running?

No. If you have Zabbix, LibreNMS, PRTG, Cacti or UISP in place, we work inside it and improve it. Most networks already have monitoring. What they don't have is someone watching it at 3am and thresholds that mean anything.

How do you get into our network?

Read-only SNMP v3 and a restricted API account, over a VPN you control, from IP addresses you whitelist. Response plans need write access, and we scope that separately and explicitly with you. Every action is logged, and access is revocable by you in a single step.

What counts as a device?

Anything we poll and alert on — an AP, a radio on a critical link, a switch, a router, a rectifier, a UPS. Subscriber CPE in bulk is priced separately and much lower. Add it to the count and we'll adjust on your first invoice if the sweep shows something different.

You're based outside the US. Is that a problem?

It's the reason the coverage works. Our team is in South Africa, which puts your overnight window in our working day. We invoice in USD, we're reachable on a US number throughout the covered hours, and everything we touch is under a signed access agreement and NDA.

Can we pay by invoice or ACH instead of card?

Yes. Annual plans and anything over 500 devices can be invoiced on net 30 with ACH or wire. Use the invoice request form below and you'll have terms, banking details and a completed W-8BEN-E the same day — that's the form your accounts team needs from a non-US supplier, in place of a W-9.

Talk first, or bill by invoice

Not ready to check out? Send a device list.

A CSV, a config export, or a rough count per site is enough. Twenty minutes on a call and you'll have a firm number, payment terms and a start date.

udisp.networks@gmail.com
+1 (000) 000-0000

We reply within one business day. Nothing is shared, and we don't add you to a list.