🧞 Unleashing Full-Stack Parallelism with Pharos

Making blockchain scalable again

Nov 25, 2024

Gm and happy Monday frens🖖

One of Web3's most famous problems is the 'blockchain trilemma' - blockchain needs to be secure, scalable and decentralized, but focusing on optimizing one aspect often comes at the expense of the others, making it difficult to achieve all three simultaneously.

Pharos, a new EVM-compatible Layer 1 blockchain, plans to change blockchain scalability, aiming to support over a billion users by integrating fintech services with the Web3 projects.

Scroll down to see:
📊 Solving scalability (finally)
🧵 Best Twitter threads/longreads of the week
📰 Neeeews

🛠️ Deep-dive: making blockchain scalable again

Pharos' approach relies on full-stack parallel technologies, which optimize three crucial aspects of blockchain infrastructure: consensus, execution, and storage.

Keynote from Join Wishlonger, Co-Founder @ Pharos, from L2con in Bangkok. Watch the full presentation here.

Challenges in blockchain scalability

Blockchain scalability presents unique challenges, including the need for extreme performance to handle billions of users and massive data volumes. Traditional systems face bottlenecks such as slow storage performance, state bloat due to redundant data storage, and inefficiencies in transaction processing.

For example, Ethereum's archive nodes require up to 19 terabytes of storage, and 52% of transaction processing time is spent on I/O operations. Pharos addresses these challenges through innovative solutions in storage, execution, and consensus.

New storage type

The Pharos Store, its blockchain-native storage system, overcomes traditional limitations through several key innovations:

First, it integrates Merkle tree nodes directly into the database, eliminating the long I/O paths that slow down transaction processing.

Second, its version addressing technique indexes data by block number, improving disk alignment and efficiency.

Finally, advanced data encoding reduces redundancy by storing only modified parts of data.

As a result, Pharos achieves 60x higher I/O throughput, 10x faster state synchronization, and an 80% reduction in storage costs. Even under stress, with 1.6 billion addresses and over 100 terabytes of data, the system maintains 4,000 TPS.

Redefining execution with Dora Compiler and VM

Pharos also redefines blockchain execution with its Dora Compiler and Dora VM. These tools introduce a unified Intermediate Representation (IR) that compiles both EVM and Wasm bytecode, optimizing for both performance and compatibility.

The system enables precise transaction-level and block-level parallelism, maximizing CPU resource utilization. Compared to existing execution engines, Dora achieves 40x better performance than REVM and is twice as fast as RGMC.

This architecture also ensures seamless interoperability between Solidity-based and Wasm-based smart contracts, making application development easier.

Consensus algorithms by Pharos

On the consensus front, Pharos employs an asynchronous Byzantine Fault Tolerance (BFT) algorithm. This approach allows all validators to propose blocks simultaneously, maximizing network bandwidth and eliminating single points of failure.

Pharos’s consensus algorithm significantly outperforms alternatives, achieving 10x higher throughput than HotStuff and 3x higher throughput than Tusk. These results, validated through publication at the SOSP conference, underscore the robustness and efficiency of the system.