How to design a resilient multi-WAN network for a yacht
Resilience comes from sound decisions about link roles, health checks and traffic policy. Buying more terminals does not provide it on its own. This guide sets out the architecture, the trade-offs and the tests that matter.
A resilient multi-WAN network for a yacht combines LEO satellite (such as Starlink Maritime), geostationary VSAT and 4G/5G cellular behind a single router. That router steers each class of traffic to the most suitable link and fails over automatically when a link degrades. Most of the design work is not in the hardware. It is in deciding what role each link plays, how its health is measured, and which onboard networks (guest, crew and vessel) may use which link.
A yacht's position, heading and surroundings change constantly. A link that is excellent at anchor may be obstructed alongside a tall quay, out of coverage mid-passage or unaffordable on a roaming tariff. A well-designed multi-WAN network expects these changes and handles them without anyone on board needing to intervene.
Give every link a clear role
Each connectivity type behaves differently. Its role should follow from those characteristics, not from which terminal was fitted most recently.
- LEO satellite (e.g. Starlink Maritime): low latency and high throughput across much of the world's sea area. It suits interactive traffic such as video calls and remote desktop work. It is sensitive to obstructions from masts, radomes and superstructure, and coverage and service terms vary by region and plan.
- Geostationary VSAT: a mature service with contracted bandwidth and established support arrangements. Because the satellite sits roughly 36,000 km above the equator, latency is high, and that is a fixed consequence of the distance involved. VSAT works well as a resilient backbone and for traffic where predictability matters more than responsiveness. Depending on frequency band, heavy rain can affect it.
- 4G/5G cellular: often the fastest and most economical option in port and close to shore. Coverage falls away quickly offshore, and roaming charges or data caps can make unmanaged use expensive.
Avoid setting one global order of preference. Define a preference order per traffic class instead. Guest video calls may prefer LEO, then cellular, then VSAT. Critical vessel traffic may be allowed on every link, including those reserved for emergencies. Bulk transfers may be barred from metered links altogether.
Health checks that reflect real performance
A link showing as up on the router does not mean it is carrying traffic properly. A terminal can hold a carrier while passing almost nothing, or have working DNS but a congested path beyond it. Failover is only as good as the checks that trigger it.
- Probe several independent targets through each link, so that one unreachable server does not condemn a working connection.
- Measure latency, jitter and packet loss, not only reachability, and set thresholds per link. A latency figure that is normal for VSAT would indicate a fault on LEO.
- Add hysteresis. Require several consecutive failures before failing over, and a sustained period of good results before failing back. Otherwise brief LEO obstructions during a turn will cause the network to flap between links.
- Use terminal status data where available. Many modems and terminals report obstruction, signal or service state. This adds a useful second signal alongside active probing.
- Check DNS resolution separately, as DNS failures often look like a dead link to users.
Tuning is a balance. Aggressive thresholds react quickly but cause unnecessary switching. Conservative ones are stable but leave users on a poor link for longer. The right setting depends on the vessel's operating pattern, and it should be revisited after the first season.
Policy routing per network: guest, crew and vessel
Segmenting the onboard network into separate VLANs and SSIDs is the foundation. It allows each population to be routed, prioritised and secured differently. Firewalling between segments should be the default, with only the flows that are genuinely needed permitted across them.
Vessel operations
This covers navigation and chart updates, monitoring, remote support access and ship systems. It should have the highest priority and the widest choice of links. If only one constrained link remains, this traffic should still get through.
Guests and owner
Guests and the owner expect connectivity that simply works. Route latency-sensitive traffic such as calls and video conferencing to the lowest-latency healthy link. Give this network priority within each link's quality-of-service rules, so a crew download does not degrade a guest's call.
Crew welfare
Crew connectivity matters for retention and wellbeing. Apply fair-share limits per device or user, and decide deliberately whether crew traffic may use metered or expensive links. Large updates and streaming can often be held to unmetered links, or scheduled for when the yacht is in port.
Application-aware rules can refine this further. For example, they can keep large cloud backups off cellular roaming, or keep a corporate VPN on the most stable available path.
Failover, sessions and when to consider bonding
When traffic moves from one link to another, its public IP address changes. Existing sessions such as VoIP calls, VPN tunnels and remote desktop connections will usually drop and need to reconnect. For general browsing this goes almost unnoticed. For a call in progress, it does not.
One option is to accept a brief interruption for most traffic. Another is to tunnel selected traffic through an aggregation point ashore. This presents a stable public address and can carry sessions across link changes, or combine links for more throughput. The cost is an extra network hop, a dependency on the shore-side service and some protocol overhead. Most vessels are best served by reserving tunnelling for the traffic that genuinely needs session continuity, rather than applying it to everything.
What to test before you rely on it
A failover design that has never been exercised is an assumption, not a design. Test at the dock, at anchor and under way, because each changes which links are viable.
- Disconnect each WAN link in turn and confirm traffic moves within the expected time, for each network segment.
- Degrade rather than disconnect: introduce loss or latency and confirm the health checks respond as intended.
- Simulate LEO obstruction and confirm hysteresis prevents flapping.
- Confirm failback once the preferred link recovers, without oscillation.
- Verify that metered-link restrictions hold, so crew bulk traffic does not land on roaming cellular.
- Test DNS, captive portals and guest Wi-Fi authentication on every link.
- Power-cycle the router and terminals, and confirm the network recovers to a known state without manual steps.
- Check that alerts reach the ETO and any remote support team when a link fails.
Record the results and the expected behaviour in a short runbook. Repeat the tests after refits, firmware updates or changes to service plans.
Monitoring and management
Resilience depends on visibility. The ETO needs a single view of every link's state and of recent failover events, along with alerts that arrive before guests start reporting problems. Management interfaces should never be exposed to the internet, and router and terminal firmware should be kept current.
Our EdgeOne platform is one way to manage this. It is supplier-neutral, so it sits above the individual links rather than tying the vessel to a single connectivity provider. It is not the only approach, and the principles in this article apply whatever management layer is chosen. Details are on our products page.
Where to start
Begin with an audit. List what terminals are fitted, what the service contracts allow, how the onboard network is segmented today and which traffic genuinely matters to the vessel's operation. A multi-WAN design should follow from that picture rather than the other way round.
We design and install maritime networks and satellite connectivity for superyachts worldwide, for both new builds and refits. If you are reviewing your vessel's connectivity, speak to our team about your operating pattern and the right architecture for it.