Encountering Byzantine General Consensus: Why Classic Multi-Paxos Is Not Used on Public Blockchains
Bitcoin is one of the most successful and widely used decentralized applications (dApps) on the Ethereum network, and has inspired numerous other projects to build their own public blockchains. Despite its success, Ethereum’s original consensus algorithm, Byzantine Fault Tolerance (BFT), is not suitable for use on traditional public blockchains. In this article, we explore why Classic Multi-Paxos, another widely used consensus algorithm, is not used on public blockchain projects.
The Problem with BFT
Byzantine General Consensus (BGC) and its variants, such as Byzantine Fault Tolerance (BFT), are designed to ensure network integrity against malicious actors. However, these algorithms have been criticized for their high energy consumption, long block times, and limited scalability. Let’s put this into perspective:
- Ethereum’s BFT algorithm requires around 50,000–100,000 TH/s (tera-hashes per second) of energy to reach consensus, which means an estimated carbon footprint of over 4 million tons per year.
- The average transaction time on the Bitcoin network is about 10 minutes, while on the Ethereum network it is about 15 seconds.
Why Multi-Paxos is not suitable
Developed by Nick Szabo in the late 1990s, Multi-Paxos is a consensus algorithm that uses multiple nodes to confirm transactions. While it offers advantages such as increased durability and scalability compared to BFT, it also has several disadvantages:
- Scalability: Multi-Paxos can lead to high transaction fees and long block times due to the need for redundant confirmation.
- Energy efficiency: The algorithm’s reliance on multiple nodes makes it more energy intensive than BFT.
- Security
: Multi-Paxos can be vulnerable to attacks because there are no guarantees of participation in the network.
Satoshi Nakamoto’s choice
So why did Satoshi Nakamoto choose Byzantine Fault Tolerance (BFT) over Multi-Paxo? The answer lies in the energy costs and scalability issues associated with BFT. At the time, Ethereum was still a relatively new project with limited resources.
However, as the project grew and became more complex, it became clear that BFT would have been prohibitively expensive. The estimated carbon footprint of a single Ethereum transaction is around 1 to 2 kg (2.2 to 4.4 lb), which equates to around 200 to 400 kg (440 to 880 lb) of carbon dioxide per year.
Why Multi-Paxos is not being implemented
Despite the energy costs and scalability limitations, Multi-Paxos is still an interesting option for some projects. However, there are several reasons why it is not being implemented:
- Limited scalability: Multi-Paxos is designed to handle a limited number of transactions per second. As the network grows, this limitation becomes increasingly apparent.
- Complexity: Multi-Paxos can be more complex to implement and maintain than BFT, especially in terms of node coordination and Byzantine identification.
- Lack of scalability: While Multi-Paxos offers better scalability than BFT, it is still not sufficient to support the large transaction volumes experienced by most public blockchains.
Conclusion
In summary, while Classic Multi-Paxos has its advantages, it is not yet suitable for traditional public blockchains due to energy costs, scalability limitations, and complexity. The high energy consumption of Ethereum’s original consensus algorithm makes it a more expensive option than BFT, which can be used as the basis for a distributed ledger.
However, the success of projects like Ethereum shows that alternative consensus algorithms can offer better sustainability, scalability, and energy efficiency. As blockchain development continues, we may see new solutions that better meet the needs of public blockchains.
