Routing
Regional footprints engineers can defend
We optimise for truthful geography: workloads land where the product path says they land, BGP and transit choices are engineered for uptime and clean UDP paths where games need them, and we avoid presenting decorative “carrier maps” that imply capacity we do not control end-to-end.
Footprint schematic
Dots represent logical presence classes (not leased capacity charts). KVM SKUs that expose geography pick the actual facility at order time.
Upstream & DDoS
Network-layer mitigation is provisioned behind our edge, with scrubbing tiers staged up to 17 TB aggregate attack-volume absorption on contracted upstream capacity. Eligible scope follows your SKU, point of presence, carrier classification, and published acceptable-use terms; pair high-risk workloads with CDN or WAF where appropriate. Delivery also relies on audited facilities and diverse transit where contracted. Read the specifics on our architecture page—especially expectations for traceroute benchmarks and volumetric floods.
Looking glass & static test IPs
We rarely publish immortal ping targets—addresses rotate across refreshes and differ by SKU, which avoids misleading benchmarks. For procurement packets, ask presales once you confirm product and geography; combine results with monitoring from your edge.
Observability
Current posture lives on our status timeline. For contractual targets, cite the relevant SLA.
Hardware lineage
Node classes—including Ryzen profiles and NVMe tiers—are described in provisioning context once you attach a SKU. Start with our datacenter briefing.
Incident hygiene
Volumetric mitigation playbook (engineering lens)
This is deliberately short so NOC narration and procurement PDFs agree. Figures stay bounded to contracted scrub tiers (currently communicated as aggregate 17 TB-class absorption)—no “infinite mitigation” folklore.
- 1. Detect & declare SOC watches upstream alarms and customer tickets in parallel. Declare an internal incident whenever scrub utilisation climbs past normal baselines—even if VMs still ping.
- 2. Sink volumetric spikes at the edge Network-layer mitigation is provisioned behind our edge, with scrubbing tiers staged up to 17 TB aggregate attack-volume absorption on contracted upstream capacity. Routed traffic resumes once upstream signals a clean verdict—document that timestamp in ticket threads.
- 3. Separate app-layer floods SYN/UDP floods behave differently than credential stuffing or Slowloris. Eligible scope follows your SKU, point of presence, carrier classification, and published acceptable-use terms; pair high-risk workloads with CDN or WAF where appropriate.
- 4. Communicate externally Broad customer impact publishes on status; KVM-specific reassurance links back to the retail mitigation briefing and changelogs for non-secret adjustments.