Architecture

One OS to rule them all.

Architecture

The Problem

Traditional infrastructure stacks are assembled from dozens of independent products. Each one has its own attack surface, its own update cycle, its own configuration language, and its own failure modes. You end up managing the glue between products more than managing your actual workloads.

Every boundary between products is a potential breach point. Every integration is a fragile handshake. Every vendor update risks breaking the one next to it.

Tophan eliminates this by being one OS with integrated Scales — each Scale replaces an entire product category, but shares the same immutable foundation, the same security model, and the same management plane.

Immutable Foundation

The base layer is read-only. Always. There is no mechanism to modify the running OS image at runtime. Configuration is declarative and applied at boot. If something goes wrong — malware, misconfiguration, cosmic rays — reboot and you’re back to a known-good state.

This isn’t a feature bolted on after the fact. The entire system is designed around immutability from the ground up.

Built for AI-Era Threats

Modern threats are automated, adaptive, and fast. Tophan’s security posture assumes the attacker is an AI:

Defence	How It Works
Immutable Base	OS image is read-only. No persistent rootkits.
Golden Images	Workloads deploy from verified, signed images.
Reboot = Reset	Any runtime compromise is wiped on restart.
Microsegmentation	Every workload is network-isolated by default.
AI-Aware Monitoring	Behavioural analysis designed for machine-speed attacks.
Automated Response	Threat response happens in milliseconds, not minutes.
Transparency	Every action is logged, auditable, and traceable.

Scales, Not Modules

Each Scale is a complete, production-ready appliance. A Scale isn’t a wrapper around someone else’s software — it’s a purpose-built system that owns its entire domain.

Scales share:

• The immutable base layer
• The security model and trust chain
• The management API (Dragon’s Eye)
• Storage and networking primitives

But each Scale owns its own lifecycle, configuration, and upgrade path. You can run one Scale or fifteen. Add them when you need them.

The Layer System

Tophan uses overlayfs composition for everything — OS, applications, and workloads.

┌─────────────────────────────┐
│       Application Layer     │  ← Workload-specific
├─────────────────────────────┤
│        Scale Layers         │  ← Feature appliances
├─────────────────────────────┤
│      Configuration Layer    │  ← Declarative config
├─────────────────────────────┤
│     Immutable Base Layer    │  ← Read-only OS image
└─────────────────────────────┘

Layers are:

• Composable — stack them in any combination
• Deduplicated — identical blocks stored once across the entire cluster
• Scannable — every layer can be integrity-checked against its signed manifest
• Versioned — roll back any layer independently

This means 100 VMs running the same base OS store that OS exactly once on disk. Updates apply to the base layer and propagate everywhere on next boot. No drift. No snowflakes.

System Stack

 ┌──────────────────────────────────┐
 │          Dragon's Eye            │  Management UI
 ├──────────────────────────────────┤
 │      tophand (Orchestrator)      │  System daemon
 ├──────┬──────┬──────┬──────┬──────┤
 │ Hyper│ Net  │ Stor │ Sec  │ ...  │  Scales
 ├──────┴──────┴──────┴──────┴──────┤
 │        Vault (Secrets)           │  TPM-backed trust
 ├──────────────────────────────────┤
 │     Immutable Base (Linux)       │  Read-only foundation
 └──────────────────────────────────┘

Everything above the base layer communicates through well-defined APIs. tophand is the system daemon that orchestrates Scale lifecycle, and Dragon’s Eye is the single interface for humans.