⚡AleoBFT: Explore Aleo`s consensus layer⚡
Aleo employs AleoBFT—a DAG-based consensus algorithm inspired by Narwhal and Bullshark—enhanced with dynamic committees and staking. Formally verified to prevent forks, AleoBFT ensures both security and correctness.
- ✔️High Throughput: Combines Narwhal’s efficient data dissemination with Bullshark’s streamlined ordering.
- 🔐Robust Security: Prevents blockchain forks via formal verification.
- 🌎Scalability: Supports decentralization while enabling privacy-preserving applications.
Narwhal is a DAG-based protocol facilitating transaction communication between nodes. It builds a DAG with certificates as vertices and links as edges, enabling high-throughput transaction processing. However, Narwhal provides only a causal order.
Bullshark complements Narwhal by interpreting its DAG to produce a single, definitive transaction order without additional message exchanges. This finalized order forms the blockchain.
By offloading data dissemination from consensus, AleoBFT reduces overhead, boosting performance while maintaining decentralization and security. It’s the backbone of Aleo’s privacy-preserving blockchain applications.
Narwhal operates by having validators run multiple workers and a primary node. Workers process transactions into batches, share them with other validators, and generate digests (certificates of availability) for the primary. Instead of including raw transactions, the primary references cryptographic hashes of these batches, reducing block size.
💫The DAG building process (and beyond)💫
The DAG is built in rounds, with validators collecting n-f certificates (a supermajority) from the previous round to ensure consensus. Validators send proposals, gather endorsements, and form quorum certificates, which are shared to advance the process. This design ensures data availability, reduces overhead, adapts to network delays, and supports scalability with robust security.
Once the DAG is built, Bullshark sequences and finalizes blocks without additional validator communication, relying on the information encoded in the DAG. Thanks to Narwhal’s design, which ensures non-equivocation (certificates created by the same party in a round are identical), the ordering logic is simplified, eliminating the need for view-change mechanisms. Validators can interpret the total order based solely on their local view of the DAG.
In every even round, a leader is deterministically chosen, and its certificate is designated as an anchor. For the anchor to be committed, it must receive at least 1+f votes from certificates in subsequent odd rounds. This ensures consistency even in an asynchronous network, as any anchor committed by one validator will always have a path in future rounds. Anchors without sufficient votes or paths to future rounds are safely skipped, streamlining the commitment process.