💡What is Avalanche Consensus?

Some background on Snow and Avalanche Consensus. Watch this video for a full explanation.

Avalanche Consensus

Watch this video for a full explanation with animations.

Landslide replaced Tendermint consensus with Avalanche consensus and runs on an Avalanche subnet. Landslide enables users to:

• run the Tendermint ABCI on an Avalanche subnet

• run any Cosmos-based dApp on an Avalanche subnet

• send assets and message across IBC to/from Avalanche to any other IBC-enabled blockchain

Overview

Avalanche consensus is a novel consensus algorithm designed for distributed ledger technology that aims to achieve high throughput, low latency, and high scalability. It is unique in that it uses a new consensus paradigm called the Avalanche family of protocols, which rely on a new approach to network sampling, voting, and confirmation.

In Avalanche consensus, nodes can propose transactions and vote on them. Unlike traditional consensus algorithms like Proof of Work or Proof of Stake, in which nodes vote on a single block, in Avalanche consensus, nodes vote on multiple transactions at the same time. This allows Avalanche to achieve high throughput, as it can process multiple transactions in parallel.

The Avalanche family of protocols is also designed to handle network partitions gracefully. In traditional consensus algorithms, a network partition can cause forks in the blockchain, which can be difficult to resolve. In Avalanche, network partitions are resolved automatically by having nodes vote on transactions independently in each partition. When the network partitions are resolved, the protocol ensures that only one transaction is accepted for each transaction ID.

Avalanche consensus is also able to achieve low latency, as transactions can be confirmed within seconds of being proposed. This is achieved through a process called sub-sampling, in which nodes sample only a small portion of the network to determine whether a transaction is valid. If a transaction is deemed valid by this small group of nodes, it is then broadcast to the rest of the network for confirmation. This allows for fast confirmation times while still maintaining the security of the system.

Sub sampling

In Avalanche consensus, the network subsampling algorithm is designed to be meta-stable, which means that it is able to maintain consensus even in the presence of temporary failures or inconsistencies in the network. This is important because it allows the protocol to continue to operate correctly even if a small subset of nodes fail or behave maliciously.

The network subsampling algorithm in Avalanche works by randomly selecting a subset of nodes from the network to participate in a transaction vote. Each selected node then votes on the transaction, and if enough nodes vote in favor of the transaction, it is considered confirmed and added to the blockchain.

In order for the network subsampling algorithm to be meta-stable, it is designed to be resilient to a variety of network failures and inconsistencies. For example, if a node fails to respond to a vote request, the algorithm is designed to select a replacement node to vote in its place. Similarly, if there are inconsistencies in the votes received from the selected nodes, the algorithm is designed to use a fallback mechanism to resolve the inconsistency and ensure consensus.

The meta-stability of the network subsampling algorithm is important because it ensures that the consensus protocol can continue to operate correctly even in the face of temporary network failures or inconsistencies. This makes Avalanche consensus more robust and reliable than traditional consensus algorithms, which may be vulnerable to these types of failures.

Overall, the meta-stability of the network subsampling algorithm is a key feature of Avalanche consensus that helps to ensure the security and reliability of the protocol.

Faster Finality

Time to finality is faster on Avalanche consensus compared to Tendermint consensus due to the probabilistic nature of the Avalanche consensus algorithm. In probabilistic consensus, transactions are not required to be confirmed by every node in the network, but rather by a statistically significant subset of nodes. This allows transactions to be confirmed more quickly, as they do not have to wait for every node in the network to confirm them.

In contrast, Tendermint consensus is a deterministic consensus algorithm in which every node in the network must confirm each transaction before it can be added to the blockchain. This can lead to longer confirmation times, as the confirmation process can be bottlenecked by slow or faulty nodes.

The probabilistic nature of Avalanche consensus allows it to achieve faster time to finality by using a process called sub-sampling. In this process, a small subset of nodes is selected to confirm each transaction, and if the transaction is confirmed by this subset, it is then broadcast to the rest of the network for confirmation. This allows transactions to be confirmed much more quickly than in deterministic consensus algorithms like Tendermint.

Additionally, the sub-sampling process in Avalanche consensus is designed to be meta-stable, which means that it is able to maintain consensus even in the presence of temporary failures or inconsistencies in the network. This further improves the speed and efficiency of the confirmation process.

Overall, the probabilistic nature of Avalanche consensus, combined with its sub-sampling process and meta-stability, allows it to achieve faster time to finality than deterministic consensus algorithms like Tendermint.

Video overview

Got 2 minutes? Check out a video overview of Dr. Emin Gün Sirer explaining Avalanche consensus: