There is a question worth asking any satellite provider before you look at their pricing: where is the hub? Not the company’s head office — the teleport, the physical ground station your terminal’s signal actually lands on. For most satellite service sold into Somalia, the answer is a facility in Europe or the Gulf. That is a defensible engineering choice in some circumstances, and a quietly expensive one in most.
The unavoidable part
A geostationary satellite sits roughly 35,786 km above the equator. Light covers that in about 119 milliseconds. Up and down again is roughly 239 ms, so a single satellite hop costs you close to a quarter of a second each way — around 480 to 600 ms for a round trip once equipment processing is counted.
Nothing about the choice of teleport changes that. It is geometry, and every GEO satellite service everywhere in the world pays it. Any provider suggesting otherwise is selling something other than GEO.
The avoidable part
What the teleport’s location decides is what happens after the signal comes down.
If the hub is in Mogadishu, traffic lands in Somalia and hands onto terrestrial capacity here. A request to a service hosted in country is answered without leaving the country. A request to the wider internet takes the subsea route out, like anyone else’s.
If the hub is in Frankfurt, every packet from your terminal arrives in Frankfurt first. For content hosted in Europe, fine — that is where it was going anyway. For anything regional, the packet has now travelled to another continent and must come back: add the terrestrial round trip, which over subsea routes between Europe and the Horn is commonly in the region of 120 to 170 ms depending on path. You have paid the satellite hop and a continental detour to reach something that may be a few kilometres from the dish.
Where it gets worse
Two situations turn an annoyance into a problem.
Site-to-site traffic. If two of your sites both connect by satellite through a foreign hub and need to talk to each other, the traffic makes two satellite hops — up, down at the hub, back up, down again. That is close to a second of round-trip latency before any application does any work. Voice becomes difficult. Anything interactive becomes unpleasant.
Chatty applications. Protocols that require several exchanges before delivering anything multiply whatever the round trip is. A database query, a login handshake, a poorly tuned VPN — each adds a full round trip. At 600 ms the application feels slow. At 900 ms people stop using it and find a workaround, which is usually worse.
The part that is not about latency at all
Latency is measurable, which makes it the easy argument. The harder one is accountability.
When the hub is abroad, your service depends on a facility your provider may not own, in a country where nobody you can reach has physical access, operated by a company you have no contract with. When something degrades, the diagnosis crosses a border and so does the fix. The local provider’s honest position is often that they have raised a ticket and are waiting — which is not negligence on their part, it is the architecture.
When the hub is in the same country as the customer, and on the same compound as the cable landing station, that entire class of conversation disappears. The engineer who can look at the equipment is a drive away. The handoff from satellite to subsea capacity happens inside one building under one operations team.
When a foreign hub is genuinely the right answer
We would rather say this than have it discovered later. A hub abroad makes sense when almost all of your traffic terminates in that region anyway — a European organisation whose applications, users and back-office all sit in Europe may see no benefit from landing in Mogadishu first. It can also make sense where a provider’s space segment, licensing or commercial arrangement only supports that architecture.
What is not defensible is choosing it by default, without anyone having asked where the traffic is actually going.
Three questions before you sign
- Where is the teleport, physically? City and country. If the answer is a company name, ask again.
- Where does my traffic reach the internet? That is the point that decides your latency to everything.
- If two of my sites talk to each other, how many satellite hops is that? The answer should be one. If it is two, ask what that does to your applications.
None of this requires trusting a vendor. Latency is measurable from the terminal on day one, and the honest providers will tell you the number before you buy rather than after.
