Mahi-Mahi is a high-throughput asynchronous consensus protocol designed to provide efficient and scalable distributed consensus with low-latency. It achieves consensus in 4 network hops and utilizes a Directed Acyclic Graph (DAG) structure, offering superior performance in decentralized systems.
- 4 Network Hops: Asynchronous consensus mechanism with minimal communication overhead.
- DAG-based Protocol: Efficient management of concurrent requests.
- High Throughput: Capable of sustaining up to 350,000 requests per second (512B request size).
- Latency Improvement: upto 40% reduction in latency compared to state-of-the-art protocols.
Mahi-Mahi offers significant improvements over existing consensus algorithms:
- upto 40% improvement in latency.
- 350k req/s throughput with a request size of 512B with 50 validators in the wide-area.
To run Mahi-Mahi in a single machine:
/bin/bash scripts/dryrun.sh
Instructions for setting up and running the experiments in a distributed setup can be found in the experiments directory.
This software is licensed as Apache 2.0.