Service 01

Baremetal

For game servers, trading, and latency-sensitive workloads.

Dedicated single-tenant servers in our Ashburn footprint — Equinix DC3 and CoreSite VA1 — on a network we run end to end, from the switch port to the dark fiber between the buildings.

from $229/mo entry price
3 HA regions
2 × 100 Gbps max network
unmetered CDN egress

Dedicated servers

CPU Cores Memory Storage Network Price
Ryzen 9 9950X 16C / 32T @ 5.7 GHz 64 GB DDR5 2 × 1 TB 2 × 40 Gbps $229/mo
Ryzen 9 9950X 16C / 32T @ 5.7 GHz 128 GB DDR5 2 × 2 TB 2 × 40 Gbps $299/mo
Ryzen 9 9950X 16C / 32T @ 5.7 GHz 192 GB DDR5 2 × 2 TB 2 × 40 Gbps $449/mo
Ryzen 9 9950X 16C / 32T @ 5.7 GHz 256 GB DDR5 2 × 4 TB 2 × 40 Gbps $549/mo
EPYC 9375F 32C / 64T @ 4.8 GHz 384 GB DDR5 2 × 4 TB 2 × 100 Gbps $1,299/mo
EPYC 9575F 64C / 128T @ 5.0 GHz 512 GB DDR5 4 × 4 TB 2 × 100 Gbps $1,999/mo

// Two platforms, six sizes; custom builds on request. Unmetered for CDN and egress-heavy workloads — no per-gigabyte meter.

DDoS protection

The workloads we host get attacked — that is the business. Protection sits in front of every port: traffic is filtered on our network before it reaches your machine, and your game or app keeps answering while the attack burns out upstream.

  • Filtering upstream of your port, on our own network — not a third-party detour.
  • Tuned for the traffic we host: game protocols, market data, web.
  • No panic null-routes: mitigation is the default response, not disconnection.
  • TODO(owner): mitigation capacity, method, and always-on vs. on-demand.

HA regions

Three regions on one backbone: two buildings in Ashburn, one in New York, meshed with our own transport. Spread machines across regions and a building failure is a failover, not an outage.

Ashburn DC3 · VA1 New York NY4 EXA Express far end TODO(owner) wavelength · TODO(owner) Fig. 1 — Overview: all fiber, Ashburn ↔ New York. Dashed: wavelength.
Ashburn 1 · DC3 Equinix DC3 Ashburn 2 · VA1 CoreSite VA1 Washington λ to NY4 dark fiber ×2 — physically separate routes, no shared conduit · TODO(owner): duct detail Fig. 2 — Zoomed: Ashburn metro, dark fiber ×2 on separate routes.

The network

Between Equinix DC3 and CoreSite VA1 we run dark fiber on two physically separate routes that share no conduit. Equinix NY4 is reached by wavelength from CoreSite VA1. At the edge, transit is blended and routes are steered for latency — around congestion, not along the cheapest path.

Blended transit

Multiple upstream carriers at the edge; no single carrier failure takes us down. TODO(owner): carrier list.

Direct peering

Peered with the networks our customers actually talk to. TODO(owner): IX presence and peering policy.

Latency-first routing

Routes measured from inside the network and steered around congestion — your players don't wait for a carrier ticket.

TODO(owner): ASN · PeeringDB entry

see the full network →

Uptime & SLA

Uptime is an engineering output, not a marketing number.

  • Written SLA with service credits. TODO(owner): uptime percentage and credit schedule.
  • Two physically diverse fiber routes in Ashburn — no shared conduit.
  • Three HA regions to spread against building failure.
  • Single-tenant machines — nothing between you and the hardware.
  • TODO(owner): status page and maintenance-notice policy.

Included

  • Single-tenant hardware. No neighbors, no hypervisor between you and the machine.
  • Deployed in Equinix DC3 or CoreSite VA1, cross-connected to our backbone.
  • Sized for latency-sensitive work: game servers, trading, real-time media.
  • Four standard SKUs, one platform — custom builds on request.
  • DDoS mitigation in front of the port. TODO(owner): capacity and method.
  • TODO(owner): provisioning lead time and IP allocation.

from $229/mo

request a quote →

TODO(owner): approve copy

← index