Documentation
Tune path behavior
By default, VIGIL-MESH connects two members immediately through a blind relay, then races toward a direct path and migrates to it without interruption as soon as it obtains one. Profiles let you adjust this behavior for a given network: prefer the direct path, keep a relay as a warm standby, or force a transport that crosses UDP-hostile networks. This page describes the path and availability profiles, and indicates when to choose what.
Path profiles
A path profile expresses the network's preference about how to transport flows between members. It does not invent new paths: it steers the choice among the options the mesh knows how to establish — a direct path between peers, transit through a blind relay, or a fallback transport when UDP is blocked.
Direct
Favors the direct path between peers as soon as it is established. The relay only serves to bootstrap the connection, then the session migrates to it. This is the profile targeted by default: latency and throughput as close as possible to the network path.
Resilient (relay as warm standby)
Uses the direct path when it is available, but maintains in parallel a relay ready to take over. If the direct path degrades or drops, the switchover to the relay is immediate rather than rebuilt cold.
Tcp-only
Forces a transport that gets through where UDP is filtered. On enterprise or public networks that block UDP, this profile favors a TCP path to remain reachable, at the cost of a less optimal transport than direct UDP.
These profiles rest on the mesh's establishment mechanism: immediate connection via relay, race toward the direct path, session migration without interruption. How these paths work is described in detail in Paths, relays and session migration.
Availability profiles
Where the path profile says how to transport, the availability profile says with what demand for continuity. It governs the effort the network devotes to keeping sessions alive and reacting quickly when the path changes: liveness probe frequency, eagerness to migrate, maintenance of a standby path.
- A continuity-oriented profile watches the paths more closely and migrates earlier: it aims to make micro-outages disappear, at the cost of a bit more control traffic and maintained standbys.
- A more frugal profile probes less often and leaves more latitude before switching over: suited to tolerant flows and to links you want to spare (metered mobile, sensors).
- The choice is made per network: a teleoperation network and a telemetry network do not have the same availability expectations.
When to use what
The best starting point is to leave the default behavior — direct targeted, relay to bootstrap, migration without interruption — and only deviate for an identified reason. The table below summarizes the typical cases.
| Situation | Path profile | Availability |
|---|---|---|
| General use, shortest latency | Direct | Balanced default |
| Critical flow that must not stutter | Resilient (relay as warm standby) | Continuity-oriented |
| Enterprise or public network that blocks UDP | Tcp-only | According to the flow's criticality |
| Tolerant telemetry, links to spare | Direct | Frugal |
- 1Start from the defaultDirect targeted, blind relay for the immediate connection, migration without interruption. It is the right choice for the majority of networks.
- 2Identify the real constraintUDP blocked? A flow that tolerates no micro-outage? A mobile link to spare? Every non-standard profile answers a precise constraint.
- 3Set path and availability per networkApply the profile to the network concerned, without generalizing to the whole workspace a setting motivated by a particular case.