How to assign votes in a distributed system
Journal of the ACM (JACM)
Implementing fault-tolerant services using the state machine approach: a tutorial
ACM Computing Surveys (CSUR)
High-Availability Computer Systems
Computer
Unreliable failure detectors for reliable distributed systems
Journal of the ACM (JACM)
The weakest failure detector for solving consensus
Journal of the ACM (JACM)
STOC '97 Proceedings of the twenty-ninth annual ACM symposium on Theory of computing
PODC '97 Proceedings of the sixteenth annual ACM symposium on Principles of distributed computing
The Byzantine Generals Problem
ACM Transactions on Programming Languages and Systems (TOPLAS)
Practical byzantine fault tolerance and proactive recovery
ACM Transactions on Computer Systems (TOCS)
Optimal Primary-Backup Protocols
WDAG '92 Proceedings of the 6th International Workshop on Distributed Algorithms
Byzantine Agreement Secure against General Adversaries in the Dual Failure Model
Proceedings of the 13th International Symposium on Distributed Computing
Synchronous Consensus for Dependent Process Failures
ICDCS '03 Proceedings of the 23rd International Conference on Distributed Computing Systems
A Generic Framework for Indulgent Consensus
ICDCS '03 Proceedings of the 23rd International Conference on Distributed Computing Systems
BASE: Using abstraction to improve fault tolerance
ACM Transactions on Computer Systems (TOCS)
Optimal early stopping uniform consensus in synchronous systems with process omission failures
Proceedings of the sixteenth annual ACM symposium on Parallelism in algorithms and architectures
DSN '05 Proceedings of the 2005 International Conference on Dependable Systems and Networks
Optimizing Threshold Protocols in Adversarial Structures
DISC '08 Proceedings of the 22nd international symposium on Distributed Computing
DISC'05 Proceedings of the 19th international conference on Distributed Computing
Hi-index | 0.00 |
To establish lower bounds on the amount of replication, there is a common partition argument used to construct indistinguishable executions such that one violates some property of interest. This violation leads to the conclusion that the lower bound on process replication is of the form $n \lfloor kt/b \rfloor$, where t is the maximum number of process failures in any of these executions and k, b are positive integers. In this paper, we show how this argument can be extended to give bounds on replication when failures are dependent. We express these bounds in terms of our model of cores and survivors sets using set properties instead of predicates of the form $n \lfloor kt/b \rfloor$. We give two different properties that express the same requirement for k 1 and b=1. One property comes directly from the argument, and the other is more useful when designing an algorithm that takes advantage of dependent failures. We also consider a somewhat unusual replication bound of $n \lfloor 3t/2 \rfloor$ that arises in the Leader Election problem for synchronous receive-omission failures. We generalize the replication bound for dependent failures, and develop an algorithm that shows that this generalized replication bound is tight.