Back to blog
Fast hardware is wasted behind a slow network. That is why we treat the network as a product of its own: designed, operated and debugged by our own engineers, not rented as a black box. Here is what it looks like today.
Three locations, one design
We run infrastructure in three European facilities:
- Amsterdam, in Qupra DC1, with a direct presence at Nikhef, one of the most connected internet hubs in Europe;
- Eygelshoven, in the Netherlands, close to the German and Belgian borders;
- Frankfurt, in iNTERWERK, a carrier-neutral facility powered by renewable energy and connected to DE-CIX.
Every location follows the same standards: at least two transit providers, redundant uplinks and in-house DDoS mitigation at the edge.
Why peering matters more than bandwidth
Bandwidth is how wide the road is. Peering decides how long the road is. When we peer directly with a network, for example a large ISP or a cloud provider, traffic between us goes straight from one network to the other instead of crossing one or more transit providers.
Fewer hops mean lower latency and fewer places where things can break. In Amsterdam we keep more than 600 peering sessions across four internet exchanges, so most European eyeballs are one hop away.
10 Gb/s on every plan
Every server, from the smallest VPS to the largest dedicated machine, is connected to a 10 Gb/s port. We do not sell faster ports as an upgrade, because the port is rarely what you pay for: it is the headroom that keeps your backups, deploys and traffic spikes from colliding.
Want to see it for yourself? Our looking glass lets you run ping, traceroute and download tests from each location before you buy.
# From your machine, compare the path to two of our locations
mtr -rwc 50 <amsterdam-test-ip>
mtr -rwc 50 <frankfurt-test-ip>
What happens when a provider fails
Each location has multiple transit providers announcing our prefixes. If one of them has an outage, BGP withdraws the broken routes and traffic shifts to the remaining paths within seconds. You might see a single retransmitted packet; you will not see downtime.
IPv6 everywhere
All locations are native dual stack. Every VPS ships with a routed /64 IPv6 block at no extra cost. Some plans are IPv6-only to keep prices down, and extra IPv4 addresses can be added in the cart when your application still needs them.
| Location | Max capacity | IXPs | Peering sessions |
|---|---|---|---|
| Amsterdam | 400 Gb/s | 4 | 600+ |
| Eygelshoven | 100 Gb/s | 2 | 100 |
| Frankfurt | 100 Gb/s | 1 | 50+ |
What is next
More peering is always the plan: every new session shortens a path for someone. If your users sit behind a network we do not peer with yet, tell us. Requests from customers have already brought several new sessions online this year.
