What we learn running ISP infrastructure
Field notes on provisioning pipelines, PPPoE at scale, captive portals and billing that reconciles itself — written by the people who keep the panels green.
Fifteen steps to a live tenant: anatomy of our provisioning pipeline
Every step is idempotent, timed to the millisecond and reversible on failure. Here's why a job chain beats a monolithic install script every single time.
PPPoE at 10,000 subscribers: what breaks first
RADIUS timeouts, stale sessions and the OLT port flaps that hide behind them — the failure modes we monitor before your NOC does.
Dunning that doesn't lose customers
Grace periods, partial payments and automatic reconciliation — how to collect more by suspending less, straight from our billing engine logs.
Captive portals that convert: vouchers, sessions and the first 30 seconds
A subscriber decides whether to pay in half a minute. What your portal needs to show — and what it must never get in the way of.
Letting an AI read the OLT: guardrails for network-aware support
Pulse runs line checks and drafts ticket replies — with a prepaid wallet, a hard stop at $0 and an audit trail on every action. Here's the setup.
What 99.99% actually costs: our health-check grid, measured
Every host and panel probed, every blip logged. A look at the monitoring stack behind the number on our homepage.
Reading about it is fine. Running it is better.
Start a free pilot and watch your own tenant provision in minutes — or book a demo and we'll walk the NOC with you live.