Background

Traditional Byzantine Fault Tolerant (BFT) proof-of-stake (PoS) consensus mechanisms, widely used in many successful blockchain projects such as Binance Chain, Terra, Oasis Labs, and Celestia, have significantly improved scalability and energy efficiency over proof-of-work systems. However, these systems face several critical limitations that hinder their overall performance and scalability.

1. Leader Bottleneck: In leader-based consensus protocols, a single nominated leader is responsible for collecting transactions, proposing blocks, and driving the consensus process. This role creates a bottleneck, as the leader's resources (storage, CPU, and bandwidth) are heavily utilized, while other validators remain underutilized. This imbalance limits the system's ability to scale efficiently, as the leader's capacity becomes the primary constraint on performance.

2. Redundant Communication: Traditional BFT consensus protocols necessitate multiple rounds of communication among validators to reach consensus on the same set of transactions. Each round involves broadcasting messages to all validators, resulting in redundant communication that increases overall message complexity and network load. This redundancy not only reduces efficiency but also lengthens block time.

3. Transaction Data Sharing: In these consensus mechanisms, transaction data is shared among validators as part of the consensus process. This approach leads to inefficiencies and increased overhead, as the same data must be transmitted and processed multiple times. The repeated transmission of transaction data contributes to higher network congestion and resource consumption.

The aforementioned limitations result in significant constraints on the transactions per second (TPS) that traditional BFT-based PoS consensus protocols can achieve. Scalaris addresses these limitations by building on the innovative research of Narwhal & Bullshark, introducing a leaderless consensus mechanism that fundamentally improves upon traditional BFT-based PoS protocols.

• Zero Communication Overhead: In Scalaris, validators independently examine their local view of the Directed Acyclic Graph (DAG) to fully order all vertices without the need for additional message exchanges. This leaderless approach eliminates the communication overhead typically associated with multiple rounds of consensus communication in leader-based systems. By allowing each validator to rely solely on its local view for consensus, Scalaris enhances overall efficiency and significantly reduces latency.

• Separation of Data Propagation and Consensus: Scalaris separates the tasks of data propagation and consensus, allowing them to run in parallel. Transaction data is disseminated efficiently, while the consensus protocol focuses on ordering metadata references rather than the full transaction data. This separation enhances throughput and reduces the complexity of the consensus process.

• Metadata Consensus: By focusing on metadata rather than full transaction data, Scalaris reduces the complexity and size of messages that need to be transmitted during the consensus process. This approach significantly enhances scalability and performance by minimizing the amount of data that must be communicated and processed.

Scalaris leverages these innovations to create a high-performance, leaderless, parallel consensus framework that offers several key advantages:

• High Throughput: Scalaris achieves significantly higher throughput compared to traditional BFT-based PoS consensus protocols, as demonstrated by Narwhal's ability to handle over 150,000 transactions per second (tx/sec) with low latency.

• MEV Mitigation: Scalaris introduces a novel mechanism for mitigating Miner Extractable Value (MEV) by randomizing the order of fixed set of transactions during the commitment process. This prevents transaction ordering manipulation and ensures fair transaction processing.

Scalaris is a framework for building blockchains quickly, utilizing state-of-the-art DAG leaderless consensus protocols such as Narwhal & Bullshark, and Mysticeti. Mysticeti, in particular, offers very high throughput, up to 400,000 transactions per second (TPS), and low latency for consensus commits, around 0.5 seconds. By integrating these advanced consensus mechanisms, Scalaris sets a new standard for blockchain performance and scalability, making it an ideal choice for building next-generation decentralized applications.