Network Architecture

Overview of the ComputeNet network architecture

Draft Documentation

This documentation is under development and may be incomplete or subject to change.

Overview

This document describes the high-level architecture of the ComputeNet network, including its layers, components, and communication patterns.

Architecture Layers

ComputeNet is organized into distinct layers:

Application Layer

SDKs, APIs, and client libraries that applications use to interact with the network.

Protocol Layer

Job submission, routing, verification, and consensus protocols that coordinate network operations.

Execution Layer

Validator nodes that execute compute jobs and generate proofs within isolated runtime environments.

Network Layer

Peer-to-peer networking, gossip protocols, and data transport between network participants.

Component Diagram

Key components and their relationships:

  • Gateway Nodes — Entry points for client requests
  • Coordinator Nodes — Manage job distribution and consensus
  • Validator Nodes — Execute jobs and generate proofs
  • Archive Nodes — Store historical receipts and state
  • Relay Nodes — Facilitate cross-region communication

Draft Architecture

The architecture is still being refined. Some components may be consolidated or restructured before mainnet.

Data Flow

A typical job flows through the network as follows:

  1. Client submits job to a gateway node
  2. Gateway validates and forwards to coordinator
  3. Coordinator selects validators and assigns job
  4. Primary validator executes and generates proof
  5. Secondary validators verify and attest
  6. Coordinator aggregates attestations
  7. Result and receipt returned to client
  8. Receipt archived for future verification

Fault Tolerance

The network is designed for resilience:

  • No single point of failure in the data path
  • Automatic failover for validator failures
  • Redundant archive storage across regions
  • Graceful degradation under partial network partition