Core Concepts
Asymmetric Byzantine quorum systems model subjective trust assumptions, where each process can make its own assumptions about which other processes may fail. This generalizes the traditional symmetric trust model and enables more flexible and realistic trust management in distributed systems.
Abstract
The paper introduces asymmetric Byzantine quorum systems as an extension of the standard symmetric Byzantine quorum systems. In the asymmetric model, every process can specify its own trust assumptions about which other processes may fail, rather than having a global trust assumption.
The key contributions are:
Formal definition of asymmetric Byzantine quorum systems and the necessary B3 condition for their existence.
Protocols for implementing shared memory (single-writer multi-reader registers) and broadcast primitives (consistent and reliable broadcast) using asymmetric quorum systems.
Discussion of the relationship between asymmetric quorum systems and the federated Byzantine quorum systems used in blockchain protocols like Ripple and Stellar, highlighting the advantages of the asymmetric model.
The protocols for shared memory and broadcast generalize the classic algorithms in the literature to the asymmetric trust setting, ensuring safety and liveness guarantees for "wise" processes that consider the actual faults, but not necessarily for "naive" processes that have incorrect trust assumptions.