Most network teams still watch their network through a keyhole: a wall of green up/down lights and SNMP counters. That’s fine until the day something breaks that up/down simply can’t see — and then the investigation begins.
The faults up/down can’t show you
Three of the most common “everything’s green but it’s broken” incidents:
- A BGP path shifts. Traffic quietly starts egressing via a backup provider because a
local-prefchanged or anas-pathlengthened after a peer reset. Latency climbs; nothing is “down.” - A multicast tree thins. A market-data or SCADA feed loses a branch — the
(S,G)mroute exists at the source but a downstream router’s outgoing-interface list is empty. Receivers go silent while every interface reports “up.” - An MPLS/LDP session drops. A customer VPN site goes dark even though the IP layer looks perfectly healthy.
None of these show up as a red light, because nothing is, technically, down.
Depth means protocol-level state, kept as history
Phantom tracks the state that explains these outages — BGP path attributes, multicast trees and reverse-path health, VRF route-targets, MPLS labels and label-distribution sessions, routing stability — and keeps a timestamped history of it. So the question stops being “is it up?” and becomes:
What changed, and when?
That’s a query, not a war room.
Why history is the unlock
A single snapshot tells you the current state. A time-series lets you diff any two points and see the exact moment a next-hop changed, a prefix’s attributes shifted, or a multicast branch disappeared. It’s the difference between guessing and knowing.
Want to see it on your own BGP paths and multicast trees? Book a demo.