Client area
Choosing your server's power is something everyone knows how to do: you look at the processor, RAM, and storage. Choosing where to place it, however, is a question that is almost always neglected, even though it directly determines the speed your users experience, the quality of your interconnections, and the jurisdiction that protects your data. A poorly placed server means a slow website for half your audience, no matter how good its technical specs are.
OuiHeberg operates in four locations: Marseille, Frankfurt, Amsterdam, and New York. Each has its own unique network personality. This guide helps you choose yours, with latency benchmarks by city, concrete use cases, and a method to measure for yourself, because ultimately, the best location is the one your own tests identify.
The three criteria that really matter
Before comparing cities, let's clarify what truly matters when choosing a location.
The first and by far the most important criterion is the latency relative to your audience. Latency is the round-trip time of a request between the user and the server, expressed in milliseconds. The closer your server is, in both distance and network path, to your visitors, the more everything feels instantaneous. The rule is simple: bring the server closer to the majority of your users, not to yourself. If you are in Paris but your audience is American, a server in New York will serve your project better than the best French server.
The second criterion is the interconnection ecosystem. Not all cities are equal in terms of "network neighborhood." Some host massive Internet Exchange Points (IX) where thousands of networks, operators, and cloud providers meet. Being hosted there means benefiting from shorter and more stable paths to a multitude of destinations.
The third criterion is jurisdiction. The country where your data physically resides determines the law that applies to it. For many projects, especially those handling personal data, hosting in France or Europe is a structuring choice, which we detail in our dedicated article on GDPR and sovereignty.
Latency benchmarks by city
The table below gives approximate round-trip latency values (in milliseconds) from four reference cities to our four locations. These values are indicative: they depend on your provider and the network route on the day. Consider them as a compass, not as absolute truth, and above all, measure from your own context (see below).
| From ↓ / To → | Marseille | Frankfurt | Amsterdam | New York |
|---|---|---|---|---|
| Paris | ~12-18 ms | ~10-15 ms | ~8-14 ms | ~75-90 ms |
| Lyon | ~8-12 ms | ~12-18 ms | ~15-20 ms | ~85-95 ms |
| Montreal | ~90-105 ms | ~90-100 ms | ~85-95 ms | ~12-25 ms |
| Dakar | ~40-60 ms | ~50-70 ms | ~50-70 ms | ~90-120 ms |
What stands out: for a French audience, the three European sites are neck and neck (just a handful of milliseconds apart, imperceptible for a typical website). The real differences appear at the extremes: a North American audience benefits greatly from being served from New York, and a West African audience takes advantage of Marseille's privileged position.
Marseille: France, Africa, and the Middle East
Marseille is our French anchor, and a remarkable choice as soon as your audience extends south of Europe. The city is one of the world's top Internet hubs, a landing point for about fifteen submarine cables connecting around forty countries in the Mediterranean, Africa, and the Middle East. For a French, Southern European, or African audience, it's hard to beat.
It's also the most environmentally friendly location, thanks to its water cooling system in a former mining gallery. We tell the whole story, cables, river cooling, sovereignty, in our dedicated article: why we host our VPS in Marseille.
Choose Marseille if: your audience is in France, Southern Europe, the Maghreb, West Africa, or the Middle East; or if French data sovereignty is a criterion for you.
Frankfurt: the beating heart of European Internet
Frankfurt hosts the DE-CIX, the largest Internet Exchange Point in Europe, a node that has broken traffic records around 18 Tbps and through which a colossal share of the continent's traffic passes. In practical terms, a server in Frankfurt is extraordinarily well connected to all of Central and Eastern Europe and benefits from one of the best network neighborhoods in the world.
Choose Frankfurt if: your audience is in Germany, Austria, Switzerland (DACH region), or Central and Eastern Europe; or if you want the densest possible European connectivity for a widely distributed service.
Amsterdam: the hub of Benelux and Northern Europe
Amsterdam is home to the AMS-IX, one of the largest exchange points on the planet, and a historic ecosystem of cloud providers, CDNs, and hosts. The city is renowned for the quality and neutrality of its interconnection, making it an ideal base to serve Benelux, the United Kingdom, and Northern Europe.
Choose Amsterdam if: your audience is in the Netherlands, Belgium, the United Kingdom, or Scandinavia; or if your project relies on services (CDN, cloud) that are strongly present in this ecosystem.
New York: the gateway to North America
If a significant portion of your users is across the Atlantic, no European site can compete: crossing the ocean mechanically adds 70 to 90 ms. New York places your server as close as possible to the US East Coast and Eastern Canada, with excellent local response times.
Choose New York if: your audience is in the United States or Eastern Canada; or if you are targeting a North American clientele for whom responsiveness is crucial (SaaS, gaming, e-commerce).
Measure it yourself: the ping and mtr method
The best table in the world will never replace a test from your own network. Here’s how to get your real numbers in just a few minutes.
The simplest test, ping measures the raw latency to an address:
ping -c 20 my-server.example.com
Run it with about twenty packets (-c 20) and look at the summary line: the avg (average) value is your reference latency, and mdev indicates stability (the lower, the better).
The most informative test, mtr combines ping and traceroute: it shows each step (each router) of the network path and the latency at each, revealing where time is lost.
mtr -rwzbc 50 my-server.example.com
Here, -r produces a report, -c 50 sends 50 cycles for a reliable average, and -w widens the display. On Windows, the equivalent of mtr is tracert or the WinMTR tool.
How to interpret: compare the average latency to each location from where the majority of your audience is (not from your own location, unless your audience is local). A difference of a few milliseconds between two European sites is negligible; a difference of several dozen milliseconds, however, is noticeable. Also test at different times of day to spot any possible congestion.
In summary: choose your location based on your audience map
Remember just one idea: start from your users' geography, not your own. Mentally (or with your traffic stats) draw your audience map, identify where the majority is concentrated, then choose the location closest to that center of gravity, verifying with a ping/mtr test.
For most French-speaking and European projects, Marseille is a solid default choice, with the unique advantage of opening towards Africa. For a Central European audience, Frankfurt; for Benelux and the north, Amsterdam; for North America, New York. And nothing stops you, for a project with a global audience, from combining several locations.
Whatever your decision, our four sites are available for your VPS hosting, in Linux VPS as well as Windows VPS. The right server, in the right place, from the start.