The ISP network health audit checklist we run before touching anything

Underperforming networks generate strong opinions. Measurement replaces them with a priority list. This is the sequence we follow on every ISP health audit, and roughly the order in which it exposes root causes.

Phase 1 — Baseline before opinions

• 14 days of per-interface utilisation, errors and discards
• Continuous latency, jitter and loss probes between every POP pair
• Flow data for the top talkers and top destination ASNs
• Subscriber-side sample tests during peak and off-peak windows

Phase 2 — Routing and policy hygiene

• Full routing table dump per edge device and per session prefix counts
• Asymmetric path detection and MTU/PMTUD validation
• Prefix filter, max-prefix and RPKI validation state review
• Redistribution and route-leak audit between IGP and BGP

Phase 3 — Edge and capacity

Edge devices fail quietly. Check CPU per core under peak PPS (not bandwidth), session counts against platform limits, CGNAT port-block exhaustion, queue drops and shaper accuracy. A device at 40% bandwidth can be at 95% packets-per-second capacity.

Phase 4 — Visibility and process

• Are outages detected by monitoring or by customers?
• What percentage of alerts are actionable?
• Is there a documented escalation matrix and runbook set?
• Does IPAM, ASN and prefix documentation match reality?

Phase 5 — Prioritised remediation

Findings are scored on customer impact versus effort and risk. Quick wins go into a maintenance window within days; architectural changes get a phased plan with rollback steps. The deliverable is a document your team can execute — with or without us.

Buying more transit to fix a routing problem is the most expensive way to keep the problem.

• #Audit
• #ISP
• #Monitoring
• #Capacity