Achieving consensus in fault-tolerant distributed computer systems: protocols, lower bounds, and simulations
Several new results in the area of fault-tolerant distributed computing are given. Specific interest is consensus protocols, that is, protocols that enable correct processors to reach agreement in the presence of disruptive behavior by faulty processors. Results presented here are as follows: two new efficient agreement protocols, one randomized and one deterministic; a method for efficiently transforming a protocol that reaches agreement on a single bit into a protocol that reaches agreement on values chosen from a larger set; a general method for compiling a protocol that tolerates relatively benign processor faults into one that tolerates more serious processor faults; and a strengthening of the known lower bound on the number of rounds of communication required by consensus protocols.
- Research Organization:
- Massachusetts Inst. of Tech., Cambridge (USA)
- OSTI ID:
- 7231018
- Country of Publication:
- United States
- Language:
- English
Similar Records
Scalable distributed consensus to support MPI fault tolerance.
Scalable and Fault Tolerant Failure Detection and Consensus