Are Byzantine Failures Really Different from Crash Failures?
When considering n-process asynchronous systems, where up to t processes can fail, and communication is by read/write registers or reliable message-passing, are (from a computability point of view) Byzantine failures “different” from crash failures? This is the question addressed in this paper, which shows that the answer is “no” for systems where \(t<n/3\).
To this end, the paper presents a new distributed simulation whose core is an extended BG simulation suited to asynchronous message-passing systems. More precisely, assuming \(t<\min (n',n/3)\), it describes a signature-free algorithm that simulates a system of \(n'\) processes where up to t may crash, on top of a basic system of n processes where up to t may be Byzantine. In addition to extending (in a modular and direct way) the basic BG simulation to Byzantine message-passing systems this simulation also allows crash-tolerant algorithms, designed for asynchronous read/write systems, to be executed on top of asynchronous message-passing systems prone to Byzantine failures.
- 3.Borowsky, E., Gafni, E.: Generalized FLP impossibility results for \(t\)-resilient asynchronous computations. In: Proceedings 25th ACM STOC, pp. 91–100. ACM Press (1993)Google Scholar
- 5.Dolev, D., Gafni, E.: Some garbage in - some garbage out,: asynchronous \(t\)-Byzantine as asynchronous benign \(t\)-resilient system with fixed \(t\)-Trojan horse inputs. Tech Report, arXiv, 14 p., July 2016. arXiv:1607.01210
- 8.Imbs, D., Raynal, M., Stainer, J., Byzantine, F.: Failures to crash failures in message-passing systems: a BG simulation-based approach. Technical Report, arXiv, 27 p., October 2015. arXiv:1510.09119
- 10.Mendes, H., Tasson Ch., Herlihy, M.: Distributed computability in Byzantine asynchronous systems. In: Proceedings 46th STOC, pp. 704–713. ACM Press (2014)Google Scholar