Byzantine fault tolerance can be fast

Author

Castro, Miguel ; Liskov, Barbara

Author_Institution

Microsoft Corp., Redmond, WA, USA

fYear

2001

fDate

1-4 July 2001

Firstpage

513

Lastpage

518

Abstract

Byzantine fault tolerance is important because it can be used to implement highly-available systems that tolerate arbitrary behavior from faulty components. We present a detailed performance evaluation of BFT, a state-machine replication algorithm that tolerates Byzantine faults in asynchronous systems. Our results contradict the common belief that Byzantine fault tolerance is too slow to be used in practice, BFT performs well so that it can be used to implement real systems. We implemented a replicated NFS file system using BFT that performs 2% faster to 24% slower than production implementations of the NFS protocol that are not fault-tolerant.

Keywords

replicated databases; software fault tolerance; software performance evaluation; BFT; Byzantine fault tolerance; NFS protocol; arbitrary behavior; asynchronous systems; faulty components; highly-available systems; performance evaluation; replicated file system; state-machine replication algorithm; Availability; Computer science; Fault tolerance; Fault tolerant systems; File systems; Laboratories; Libraries; Optimization; Production systems; Protocols;

fLanguage

English

Publisher

ieee

Conference_Titel

Dependable Systems and Networks, 2001. DSN 2001. International Conference on

Conference_Location

Goteborg, Sweden

Print_ISBN

0-7695-1101-5

Type

conf

DOI

10.1109/DSN.2001.941437

Filename

941437