Everything your bare metal depends on is in our own hands — the network, the hardware, the racks in Amsterdam and the engineers who run them. We are not a datacenter operator: we run our own racks inside carrier-grade Amsterdam facilities, and everything from the rack upward is ours. That is what lets us fix things ourselves, immediately.
10 Gbit, 25 Gbit, 100 Gbit — whatever your workload asks for, over redundant connections to the major internet backbones. Low latency and reliable reachability worldwide, also at peak hours when other networks start to congest.
Unmetered uplinks are what we are known for. Capacity is bought for the traffic our customers actually push, never oversold against the hope that nobody uses it. An unmetered port here means the port, not a fair-use clause with a counter behind it.
And if the path to your users is slow, lossy or takes a strange detour, tell us. The routing is ours, so we can look at it with you and move your traffic onto a better path. Try getting that done at a provider that buys its transit as a product.
Bringing your own address space? We announce your IPv4 and IPv6 prefixes from our network, and IPv6 is supported everywhere on it.
We do not build datacenters, we choose them, and this is what we choose them on: redundant feeds with UPS and generator backup, and cooling dimensioned for continuous full load. We monitor what the facility delivers to our racks rather than taking it on trust.
The facilities we work from have biometric access control, round-the-clock surveillance and fire suppression. Inside them our racks are ours: only our own engineers have their hands on the hardware.
Links, routers and paths come in pairs, so hardware is upgraded, replaced or patched while your traffic takes the other side. Practically all of our maintenance happens without anyone going offline, and the rare exception is announced well in advance.
Power draw, temperatures, links, routing and the machines themselves are watched continuously. In most cases we are already working on a problem before a customer notices, and the engineer who sees the alert is the one who fixes it.
Power and cooling redundant in the facility, carriers and routing redundant in our own network. No single point can take your server offline, and on our side there is no third party who decides when it comes back.
We plan ahead and add capacity before it runs out, so growth is planned work instead of an emergency with a maintenance window attached to it.
Every machine we hand over sits on the same foundation.
“100 Gbit” is an abstraction until you need to move something. Drop in the amount of data you shift and see what each port speed does to the clock.
The difference between finishing before lunch and finishing tomorrow is usually not the servers. It is the pipe they sit on.
At line rate, decimal units. Real throughput depends on your protocol, your disks and the far end — but the ratio between the ports does not change.
Contact us to find out how this fits what you need to run, whether that is our hardware or your own in our racks.