Edge Computing & GeoDNS

In the centralized cloud model, data travels thousands of miles to reach a data center. For real-time applications (gaming, IoT, video streaming), this latency is unacceptable.

Cyberun Cloud is evolving from a "Centralized Federation" to a "Hyper-Distributed Mesh." Our roadmap involves deploying lightweight edge clusters closer to the end-user, unified by intelligent DNS routing.

The Edge Strategy: Lightweight K8s

Running a full-blown Kubernetes cluster requires significant resources. To conquer the edge, we utilize RKE2 (and K3s), fully CNCF-certified Kubernetes distributions designed for resource-constrained environments.

Architecture

The Core: Heavy workloads (AI training, Big Data) remain in our primary regions (Tokyo, NY, Nuremberg).
The Edge: Latency-sensitive microservices (API Gateways, Caching Layers, SSR Frontends) are deployed to Edge Nodes in varying POPs (Points of Presence).
Unified Management: Despite being lightweight, these edge clusters are registered to the Karmada control plane in Tokyo. You manage a Raspberry Pi cluster in London the exact same way you manage a bare-metal rack in New York.

Intelligent Routing (GeoDNS)

Having servers everywhere is useless if users are routed to the wrong one. We are implementing a Global Traffic Director based on GeoDNS.

How it Works

User Request: A user in Paris queries api.cyberun.cloud.
Geo-Location: Our authoritative DNS server detects the source IP originates from France.
Health Check: The system verifies the health of the nearest POP (e.g., Nuremberg).
Routing: The DNS resolves to the VIP of the Nuremberg HAProxy ingress.
Failover: If Nuremberg is down, the DNS automatically resolves to the next best location (e.g., New York) with a slightly higher RTT but guaranteed availability.

Why This Matters

Performance: Reduces Round-Trip Time (RTT) by routing users to the physically nearest entry point.
Compliance: Helps enforce "Data Residency" by ensuring traffic from specific regions stays within legal boundaries.