Skip to content
This repository has been archived by the owner. It is now read-only.
Switch branches/tags
Go to file
Cannot retrieve contributors at this time

Consensus Roadmap for Radix Public Network

Last updated: Friday, January 22, 2021

Cerberus is the consensus protocol designed for the Radix Public Network. The full Cerberus protocol will enable a parallelized network capable of practically unlimited, linearly scalable transaction throughput. Our first Radix Public Network release, Olympia (also called "RPN-1"), will not implement the full Cerberus protocol, but instead a simplified, unsharded version that will allow us to deploy a network with maximum confidence.

(An overview of development progress toward release of Olympia can be found here.)

Fully sharded Cerberus will be developed in multiple stages toward a Radix public network upgrade beyond Olympia to be announced. This roadmap briefly summarizes the primary consensus design features of these three anticipated consensus development stages.

Note that these are internal development stages – not necessarily releases intended to be released immediately to the Radix public network. In particular, State-sharded Cerberus is likely to be an internal milestone only and not released for the Radix public network. It is a necessary development stage, but is unlikely to offer any practical throughput benefit over unsharded Cerberus. Fully Sharded Cerberus is our target for a later upgrade to the Radix public network.

A roadmap for application layer functionality using the Radix Engine is independent of the consensus roadmap and is not considered here.

This roadmap is a current snapshot of thinking and is very likely to change in the course of development.

Unsharded Cerberus (in progress)

Unsharded Cerberus summary

Goal Deploy BFT 3-phase commit foundation for Cerberus
Data Structure Functionally unsharded, with Atoms grouped into global blocks and structured for later sharding
Validator Node Set Size Fixed size, minimum 100
Node Mapping All nodes mapped to all shards
Sybil Protection DPoS

State-sharded Cerberus

State-sharded Cerberus

Goal Deploy sharding of Atoms within Cerberus, but continuing to require all nodes to validate all shards
Data Structure Fully sharded to enable dependency-driven parallelization
Validator Node Set Size Fixed size, minimum 100
Node Mapping All nodes mapped to all shards
Sybil Protection DPoS

Fully Sharded Cerberus

Fully Sharded Cerberus

Goal Deploy full Cerberus with consensus parallelized across many physical nodes, achieve linear scalability
Data Structure Fully sharded
Validator Node Set Size Unlimited
Node Mapping Nodes mapped dynamically to shards for hardware parallelization while maintaining per-shard overlap for security
Sybil Protection DPoS or potential alternative if available