Sara Tucci-Piergiovanni

TERA: topic-based event routing for peer-to-peer architectures

The completely decoupled interaction model offered by the publish/subscribe communication paradigm perfectly suits the interoperability needs of todays large-scale, dynamic, peer-to-peer applications. The unmanaged environments, where these applications are expected to work, pose a series of problems (potentially wide number of partipants, low-reliability of nodes, absence of a centralized authority, etc.) that severely limit the scalability of existing approaches which were originally thought for supporting distributed applications built on the top of static and managed environments. In this paper we propose an architecture for implementing the topic-based publish/subscribe paradigm in large scale peer-to-peer systems. The architecture is based on clustering peers subscribed to the same topic. The major novelty of this architecture lies in the mechanism employed to bring events from the publisher to the cluster (namely outer-cluster routing). The evaluation shows that this mechanism for outer-cluster routing has a probability to bring events to the destination cluster very close to 1 while keeping small the involved number of out-of-cluster peers. Finally, the overall architecture is shown to be scalable along several fundamental dimensions like number of participants, subscriptions, and to exhibit a fair load distribution (load distribution closely follows the distribution of subscriptions on nodes).

Eventual Leader Election in Infinite Arrival Message-Passing System Model with Bounded Concurrency

We study the failure detection problem in a message-passing system that may dynamically change over time, so that the number of processes which make progress during a computation may grow to infinity as time tends to infinity but the number of concurrently up processes do not exceed a known bound. We first propose the specification of a new oracle, called HB*, able to give hints on which processes are making progress in the system. A possible HB* implementation is given. Then, we show how to use HB* to implement the oracle Ω that eventually identifies a unique leader in the system. To the best of our knowledge this is the first implementation of Ω running in a message passing system with infinitely many processes.

An adaptive coupling-based algorithm for internal clock synchronization of large scale dynamic systems

This paper proposes an internal clock synchronization algorithm which combines the gossip-based paradigm with a nature-inspired approach coming from the coupled oscillators phenomenon. The proposed solution allows a very large number of clocks to self-synchronize without any central control, despite node departure and arrival. This addresses the needs of an emergent class of large-scale peer-to-peer applications that have to operate without any assumptions on the underlying infrastructure. Empirical evaluation shows extremely good convergence and stability under different network settings.

Churn resilience of peer-to-peer group membership: a performance analysis

Partitioning is one of the main problems in p2p group membership. This problem rises when failures and dynamics of peer participation, or churn, occur in the overlay topology created by a group membership protocol connecting the group of peers. Solutions based on Gossip-based Group Membership (GGM) cope well with the failures while suffer from network dynamics. This paper shows a performance evaluation of SCAMP, one of the most interesting GGM protocol. The analysis points out that the probability of partitioning of the overlay topology created by SCAMP increases with the churn rate. We also compare SCAMP with DET – another membership protocol that deterministically avoids partitions of the overlay. The comparison points out an interesting trade-off between (i) reliability, in terms of guaranteeing overlay connectivity at any churn rate, and (ii) scalability in terms of creating scalable overlay topologies where latencies experienced by a peer during join and leave operations do not increase linearly with the number of peers in the group.

Brief Announcement: Eventual Leader Election in the Infinite Arrival Message-Passing System Model

We study the the failure detection problem in a message-passing system that may dynamically change over time, so that the number of processes which make progress during a computation may grow to infinity as time tends to infinity. We first propose an oracle wellsuited to such systems, called HB, by detailing its specification and a possible implementation. HB is able to eventually identify alive processes in the system. Then, we show how to use HB to implement the oracle Ω that eventually identifies an unique leader in the system. To the best of our knowledge this is the first implementation of Ω running in a system with infinitely

Communication Channel Management for Maintenance of Strong Overlay Connectivity

A fundamental problem for both structured and unstructured peer-to-peer networks is how to maintain connected the topology of a network in the presence of processes that, possibly concurrently, join and leave the network. In this paper we firstly define a model of the computation well-suited to analyze connectivity maintenance among processes carrying out a distributed computation considering unbounded concurrency and infinite participation. Secondly upon this model we provide a specification of the connectivity maintenance problem. We finally present a protocol that guarantees connectivity maintenance by arranging processes of the computation on a tree. The protocol handles both joins and leaves concurrently and actively (i.e., some piece of code is executed by a leaving/joining process interacting with its neighbors in the topology).