In this work we consider the problem of leader election, abstraction used by many distributed services to select a unique process for coordinating actions. We propose an eventual leader election algorithm for partially synchronous systems prone to concurrent crash-recovery and omission failures where any process may suffer failures forever as long as a majority of processes meet some weak connectivity and reliability conditions.

We investigate the performance of the ChandraToueg and Paxos consensus algorithms, using as case study TrustedPals, a smartcard-based framework which allows to implement security policies in distributed systems. The current architecture of TrustedPals uses the Chandra-Toueg consensus algorithm adapted to the omission failure model. In this work, we propose to alternatively use the Paxos consensus algorithm, in order to extend the applicability of the framework to the crash-recovery failure model as well. We have performed simulations to compare these two consensus algorithms under different failure scenarios. Our results show that Paxos is more efficient than Chandra-Toueg if the first process which coordinates or leads a round suffers a failure, while both algorithms perform similarly when there are no failures.

This paper presents an evaluation of three communication-efficient algorithms implementing the Omega class of failure detectors, which provides an eventual leader election functionality, in distributed systems where processes can crash and recover. Communication efficiency means that eventually only a correct process, i.e., the elected leader, keeps sending a message periodically to the rest of processes. The first algorithm relies on the use of stable storage to store the identity of the leader and an incarnation number. The second algorithm does not use stable storage, but requires a majority of correct processes. Also, it is near-communication-efficient, since besides the leader, unstable processes, i.e., those that crash and recover infinitely often, may send messages periodically before they receive a message from the leader. Finally, the third algorithm does neither use stable storage nor require a majority of correct processes, but assumes that each process has access to a nondecreasing and persistent local clock. Using the OMNeT++ network simulation framework, we evaluate the performance and the quality of service provided by these algorithms, in terms of the number of messages exchanged among processes and the capability of the failure detector to provide a single leader, respectively.

We present a modular redesign of TrustedPals, a smart card-based security framework for solving Secure Multiparty Computation (SMC). Originally, TrustedPals assumed a synchronous network setting and allowed to reduce SMC to the problem of fault-tolerant consensus among smart cards. We explore how to make TrustedPals applicable in environments with less synchrony and show how it can be used to solve asynchronous SMC. Within the redesign we investigate the problem of solving consensus in a general omission failure model augmented with failure detectors. To this end, we give novel definitions of both consensus and the class \diamond {\cal P} of failure detectors in the omission model, which we call \diamond {\cal P}({ om}), and show how to implement \diamond {\cal P}({ om}) and have consensus in such a system with very weak synchrony assumptions. The integration of failure detection and consensus into the TrustedPals framework uses tools from privacy enhancing techniques such as message padding and dummy traffic.

No abstract available.

This paper presents an evaluation of several communication-optimal algorithms implementing the diamP class of failure detectors. The first algorithm is based on a reliable broadcast primitive, involving a quadratic number of messages to manage a suspicion. The second algorithm uses exclusively one-to-one communication, involving a linear number of messages to manage a suspicion, but with a higher latency to propagate the suspicion to the rest of processes. A third algorithm reduces this latency using an additional one-to-all communication mechanism. We evaluate the quality of service provided by these algorithms, in terms of the capability of the failure detector to provide right answers and the reaction time after a failure.

No abstract available.