Zellular Docs
  • Introduction
  • Getting Started
    • The Current Landscape
    • Zellular's Solution
    • Use Cases
      • AVS Task Management
      • AVS Database Replication
    • Core Concepts
    • Performance & Scalability
  • Page
  • Technical Documentation
    • Architecture
    • Protocol
    • SDK
    • Integration
Powered by GitBook
On this page
  • Architecture Overview
  • Node Network Structure
  • Fault Detection and Recovery
  • Consensus Mechanism
  • Security Model
  1. Getting Started

Core Concepts

Architecture Overview

Zellular functions as a decentralized (BFT) message queue, enabling decentralized applications to propagate updates across their nodes while maintaining consistency and order. Zellular provides the following key features to applications using it for database replication:

  • Consistent State Management: All nodes receive and process updates in the same sequence

  • Guaranteed Message Delivery: Messages are reliably delivered to all participating nodes

  • Ordered Processing: Sequential processing ensures state consistency across the network

  • Fault Tolerance: System continues to function even if some nodes fail or misbehave

Node Network Structure

The network operates with a leader node (Sequencer) that:

  • Orders incoming transactions

  • Distributes transactions to replica nodes

  • Maintains network consistency

  • Aggregates signatures from nodes

  • Distributes consensus proofs to replicas

The network also includes replica nodes that:

  • Receive transactions from application nodes and post ordered transactions back

  • Replicate state based on Sequencer transactions

  • Validate proofs independently to ensure integrity

  • Contribute signatures for network consensus

  • Provide redundancy and enable failover

Fault Detection and Recovery

The system includes robust mechanisms for handling sequencer failures:

  • Continuous monitoring of sequencer behaviour

  • Automatic detection of malfunctions or censorship

  • Democratic challenge and replacement process

  • Seamless leadership transitions

Consensus Mechanism

Unlike traditional gossip protocols, Zellular uses an efficient leader-based two-round propagation system:

  1. First Round: Node → Sequencer

  2. Second Round: Sequencer → All Nodes

The state finalization works in the following way:

  1. Batch Collection: Transactions are grouped into batches

  2. Chaining: Each batch is linked through chaining hashes

  3. Signature Collection: Nodes sign the chaining hash

  4. Finalization: Aggregated signatures confirm finality

Zellular uses BLS signature aggregation for efficient finalization proof verification.

Security Model

Zellular leverages EigenLayer's restaking infrastructure to provide:

  • Economic security through staked ETH

  • Proven Byzantine fault tolerance

  • Decentralized node network

  • Incentive alignment for operators

PreviousAVS Database ReplicationNextPerformance & Scalability

Last updated 5 months ago