Beyond One-third Faulty Replicas in Byzantine Fault Tolerant Systems

This report describes the BFT2F algorithm implementation as a Python library, building from the popular CastroLiskov PBFT. The main focus is on investigating the system’s behavior beyond f failures, covering the, often ignored, spectrum between full correctness and arbitrary failure. Similar to PBFT, BFT2F provides liveness and consistency when up to f replicas fail. In scenarios where f < failures ≤ 2f, BFT2F confines malicious servers to specific consistency violations, providing a nuanced approach to system behavior. This report presents a practical implementation of the BFT2f algorithm, designed to enhance reliability and security in distributed systems.
This report describes the design, implementation, testing, and validation that were performed in this project and provides a general view of the application-level facts on Byzantine Fault Tolerance in distributed environments.

http://www.scs.stanford.edu/jinyuan/bft2f.pdf

Overviews and surveys are available in order to perform a quick evaluation and comparison among the protocols above:
https://arxiv.org/pdf/1707.01873.pdf
https://infoscience.epfl.ch/record/121590/files/TR-700-2009.pdf
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8014672
https://www.cs.unc.edu/~reiter/papers/1998/DC.pdf

Finally, the following readings may of interest for who is interest in understanding the many impossibility results affecting the distributed consensus:
https://www.the-paper-trail.org/post/2008-08-13-a-brief-tour-of-flp-impossibility/
https://people.eecs.berkeley.edu/~luca/cs174/byzantine.pdf