SentinelAura 🌍

Visual Geographic Arbitrage Engine for Kubernetes

Overview • Architecture • Features • Getting Started

---

Overview

SentinelAura is a high-performance Kubernetes Operator and 3D Dashboard designed to visualize and execute geographic compute arbitrage in real-time.

By tracking Spot Instance prices across 5 global infrastructure regions, SentinelAura identifies sudden cost drops (arbitrage opportunities). When a profitable margin is found, operators can trigger a 1-Click Migrate. The backend operator performs a dry-run calculation to guarantee data egress costs don't eclipse projected savings, before seamlessly scaling down the deployment in the expensive cluster and scaling it up in the cheaper one.

---

Architecture & Tech Stack

SentinelAura is built with a strict separation of concerns, heavily utilizing concurrent Go and WebGL.

⚙️ Backend (Golang Operator)

• k8s.io/client-go: Directly interfaces with the Kubernetes API. Uses a custom RegionManager to hold multiple kubernetes.Clientset instances, allowing simultaneous cross-cluster operations.
• Contextual Timeouts: Utilizes context.WithTimeout to ensure network partitions to remote clusters don't block the Goroutines handling migrations.
• Redis: Acts as a high-speed TTL cache for the fluctuating Spot API prices to prevent rate-limiting the Oracle.

🎨 Frontend (Next.js 15 & React-Three-Fiber)

• WebGL Spatial Dashboard: Renders a Points-based 3D Earth using custom ShaderMaterial for atmospheric glow.
• framer-motion-3d: Powers the physics-based camera rotation and spring transitions when regions are focused.
• Tailwind v4: Implements a strict, dark-themed "Memoria" design system alongside vibrant, gamified neon markers for immediate visual alerts.

---

Core Features

1. Multi-Cluster Orchestration

The Go backend doesn't just mock data; it maps specific Region IDs (e.g., US-East) to actual Kubernetes Contexts (ctx-us-east). The MigrationController orchestrates the scale-down/scale-up events atomically.

2. Safety DryRun Logic

Migrating compute isn't free. The Go Operator includes a safety function that calculates: (Spot Savings Over 24h) - (Egress Data Transfer Cost) = Margin. If the margin is negative, the migration is blocked with a 412 Precondition Failed error displayed instantly on the HUD.

3. Geographic Interactive Markers

The frontend dynamically converts Latitude and Longitude into 3D Vector coordinates (Math.sin(phi) * Math.cos(theta)), binding interactive HTML elements precisely to their global positions on the rotating WebGL sphere.

---

Getting Started

Prerequisites

• Docker (for Redis)
• Go 1.22+
• Node.js 20+
• (Optional) Multiple kubeconfig contexts for true multi-cluster testing.

Local Development

1. Start the Cache

docker run -p 6379:6379 -d redis

2. Run the Go Operator

cd backend
go run .
# Defaults to Mock Mode if KUBECONFIG is missing

3. Run the 3D Dashboard

cd frontend
npm install
npm run dev

Visit http://localhost:3000 to interact with the arbitrage engine.