No CPU stealing
Your vCPUs map to dedicated physical threads. No noisy neighbor stealing cycles.
Hardware configurations built for performance.
We operate multiple hardware configurations across our network. Hardware can vary between locations and nodes depending on the infrastructure available in each region, while every configuration is selected for reliable and consistent performance.
01Specifications
Hardware configurations built for performance.
We operate multiple hardware configurations across our network. Hardware can vary between locations and nodes depending on the infrastructure available in each region, while every configuration is selected for reliable and consistent performance.
Processor
AMD EPYC 7K62
High clocks, sustained boost
Memory
DDR5
Low latency channels
Storage
NVMe Gen 4
High-speed storage
Uplink
10 Gbps
High-capacity networking
Mitigation
Always-on filtering
Inline at the edge
Monitoring
Node telemetry
Continuous checks
Hardware configurations vary by location. Current specifications for each region are shown on the respective service and order pages.
02Density
Headroom is not a luxury feature.
We cap node density well below what the hardware could technically hold. Your workload gets consistent CPU time — not the leftovers after overselling.
~65%
Typical CPU utilization ceiling per host node.1:1
vCPU to physical thread ratio. No overcommit.Reserved
Memory headroom for burst workloads and OS overhead.Memory guaranteed
RAM is allocated, not ballooned. What you pay for is what you get.
Storage isolation
Each VPS and game server gets its own NVMe namespace. No noisy neighbor I/O.
Transparent metrics
Host-level utilization visible in the panel. You see what we see.
03Network
Low latency by design, not accident.
We peer directly with major transit providers and route for latency first. New locations open as demand requires — tell us where your players are.
| Location | Status |
|---|---|
| Frankfurt, Germany | Operational |
| Delhi, India | Operational |
| Mumbai, India | Operational |
Direct peering
We peer with 200+ networks at major IXPs. Fewer hops, lower latency.
Anycast ready
DDoS mitigation uses anycast to absorb attacks at the nearest scrubbing center.
10 Gbps per node
Dedicated uplink per host. No shared 1 Gbps bottlenecks between racks.
Looking glass
Public BGP looking glass and speed test endpoints per location.
Live network status and BGP looking glass available on our status page.
04Reliability
Built for uptime, operated for trust.
Infrastructure choices that reduce the chance of surprises — and honest communication when they happen.
01
Redundant power
N+1 UPS and generator at every facility. Dual PDU feeds per rack.
02
RAID storage
NVMe mirrored (RAID 1) on every node. Drive failure = zero downtime, zero data loss.
03
Automated failover
Node health monitored every 30 seconds. Unhealthy nodes drain and evacuate automatically.
04
Off-site backups
Daily backups replicated to a geographically separate facility. RPO ≤ 24h, RTO ≤ 4h.
05
Maintenance windows
Scheduled, communicated in advance. Emergency maintenance is rare and transparent.
06
Status page
Real-time component status, incident history, and scheduled maintenance calendar.
Infrastructure you can build on.
High-frequency compute, NVMe storage, 10 Gbps networking, and DDoS protection included on every plan.