Network & Datacenter
Where your server actually runs
One location, chosen on purpose: OVH US East at Vint Hill, Virginia, inside the busiest peering region on the internet. Here is the hardware, the network, and addresses you can test yourself.
The facility
Vint Hill, Virginia
Forty miles from Washington DC, inside the Ashburn interconnection corridor.
OVH US -- Vint Hill, Virginia
Every customer server and every Cloud App runs in OVH's US East facility in Vint Hill, about forty miles from Washington DC. One location, stated plainly, rather than a map of cities we do not have equipment in.
The Northern Virginia corridor
Vint Hill sits inside the Ashburn peering region, the densest interconnection market in the world. It is where the networks you actually talk to already are, which is why the numbers below look the way they do.
We own our IP addresses
Our IPv4 space is allocated to Caliber Node directly by ARIN, not rented from the datacenter. Practically: your address belongs to us, so we can move you to different hardware -- or change upstream entirely -- and your IP goes with you.
OVH builds its own
OVH designs and assembles its own servers and runs its own global fibre backbone rather than renting transit. That is the reason a small provider can sit on this kind of network at this kind of price.
DDoS filtering upstream
Volumetric attack filtering runs on OVH's network, ahead of our racks, so an attack is absorbed before it reaches the machine your server is on. It is included, not an add-on.
Nodes measured in hundreds of microseconds
Our machines sit close enough together to answer each other in about a quarter of a millisecond. Your database and your app are not in different buildings, or different racks.
The hardware
The processors your server runs on
Named, not described, and read from the machines themselves rather than typed onto this page. If a host will not tell you what the CPU is, there is usually a reason.
US East (NVMe)
AMD Ryzen 7 3800X 8-Core Processor
16 threads
US East (HDD)
AMD EPYC 7351P 16-Core Processor
32 threads
Measured, not claimed
Latency from our own racks
Taken with ping from our own equipment. Your own result will differ -- that is the point of publishing addresses you can test.
Between our own nodes
0.25 ms
node to node, same facility
To Cloudflare (1.1.1.1)
2.2 ms
public resolver, Ashburn region
To Google (8.8.8.8)
1.8 ms
public resolver, Ashburn region
Test it yourself
Addresses you can ping
Run ping, mtr or traceroute against these from wherever your users are. We would rather you measured than took our word for it.
Cloud Apps host
IPv4, our own ARIN allocation
23.136.172.7Cloud Apps host
IPv6
2604:2dc0:100:2ed6::2These answer ICMP so you can measure latency and routing. For a throughput test, ask support and we will set one up rather than leave a permanent download target on a machine serving customers.
Questions
About the location
Ready to deploy?
Spin up a production-ready server in under a minute. No contracts, no setup fees.