Transactional memory (TM) is a promising approach for designing concurrent data structures, and it is essential to develop better understanding of the formal properties that can be achieved by TM implementations. Two fundamental properties of TM implementations are disjoint-access parallelism, which is critical for their scalability, and the invisibility of read operations, which reduces memory contention.This paper proves an inherent tradeoff for implementations of transactional memories: they cannot be both disjointaccess parallel and have read-only transactions that are invisible and always terminate successfully. In fact, a lower bound of Ω(t) is proved on the number of writes needed in order to implement a read-only transaction of t items, which successfully terminates in a disjoint-access parallel TM implementation. The results assume strict serializability and thus hold under the assumption of opacity. It is shown how to extend the results to hold also for weaker consistency conditions, serializability and snapshot isolation.

International audienceIn this paper, we study the exclusive perpetual exploration problem with mobile anonymous and oblivious robots in a discrete space. Our results hold for the most generic settings: robots are asynchronous and are not given any sense of direction, so the left and right sense (i.e. chirality) is decided by the adversary that schedules robots for execution, and may change between invocations of a particular robots (as robots are oblivious). We investigate both the minimal and the maximal number of robots that are necessary and sufficient to solve the exclusive perpetual exploration problem. On the minimal side, we prove that three deterministic robots are necessary and sufficient, provided that the size n of the ring is at least 10, and show that no protocol with three robots can exclusively perpetually explore a ring of size less than 10. On the maximal side, we prove that k = n − 5 robots are necessary and sufficient to exclusively perpetually explore a ring of size n when n is co-prime with k

Abstract. In this paper, we investigate the exclusive perpetual exploration of grid shaped networks using anonymous, oblivious and fully asynchronous robots. Our results hold for robots without sense of direction (i.e. they do not agree on a common North, nor do they agree on a common left and right ; furthermore, the "North" and "left" of each robot is decided by an adversary that schedules robots for execution, and may change between invocations of particular robots). We focus on the minimal number of robots that are necessary and sufficient to solve the problem in general grids. In more details, we prove that three deterministic robots are necessary and sufficient, provided that the size of the grid is n × m with 3 ≤ n ≤ m or n = 2 and m ≥ 4. Perhaps surprisingly, and unlike results for the exploration with stop problem (where grids are "easier" to explore and stop than rings with respect to the number of robots), exclusive perpetual exploration requires as many robots in the ring as in the grid. Furthermore, we propose a classification of configurations such that the space of configurations to be checked is drastically reduced. This pre-processing lays the bases for the automated verification of our algorithm for general grids as it permits to avoid combinatorial explosion.

No abstract

This paper presents Combine, a distributed directory protocol for shared objects, designed for large-scale distributed systems. Directory protocols support move requests, allowing to write the object locally, as well as lookup requests, providing a read-only copy of the object. They have been used in distributed shared memory implementations and in data-flow implementations of distributed software transactional memory in large-scale systems. The protocol runs on an overlay tree, whose leaves are the nodes of the system; it ensures that the cost of serving a request is proportional to the cost of the shortest path between the requesting node and the serving node, in the overlay tree. The correctness of the protocol, including starvation freedom, is proved, despite asynchrony and concurrent requests. The protocol avoids race conditions by combining requests that overtake each other as they pass through the same node. Using an overlay tree with a good stretch factor yields an efficient protocol, even when requests are concurrent.

Distributed Shared Memory abstraction (DSM) is traditionally realized through a distributed memory consistency system(MCS) on top of a message passing system. In this paper we analyze the impossibility of e cient partial replication implementation of causally consistent DSM. E ciency is discussed in terms of control information that processes have to propagate to maintain consistency. We introduce the notions of share graph and hoop to model variable distribution and the concept of dependency chain to characterize processes that have to manage information about a variable even though they do not read or write that variable. Then, we weaken causal consistency to try to de ne new consistency criteria weaker enough to allow e cient partial replication implementations and strong enough to solve interesting problems. Finally, we prove that PRAM is such a criterion, and illustrate its power with the Bellman-Ford shortest path algorithm.

International audienceThe transactional approach to contention management guarantees atomicity by aborting transactions that may violate consistency. A major challenge in this approach is to schedule transactions in a manner that reduces the total time to perform all transactions (the makespan), since transactions are often aborted and restarted. The performance of a transactional scheduler can be evaluated by the ratio between its makespan and the makespan of an optimal, clairvoyant scheduler that knows the list of resource accesses that will be performed by each transaction, as well as its release time and duration. This paper studies transactional scheduling in the context of read-dominated workloads; these common workloads include read-only transactions, i.e., those that only observe data, and late-write transactions, i.e., those that update only towards the end of the transaction. We present the B IMODAL transactional scheduler, which is especially tailored to accommodate read-only transactions, without punishing transactions that write most of their duration, called early-write transactions. It is evaluated by comparison with an optimal clairvoyant scheduler; we prove that B IMODAL achieves the best competitive ratio achievable by a non-clairvoyant schedule for workloads consisting of early-write and read-only transactions. We also show that late-write transactions significantly deteriorate the competitive ratio of any non-clairvoyant scheduler, assuming it takes a conservative approach to conflicts

Considering autonomous mobile robots moving on a finite anonymous graph, this paper focuses oil the Constrained Perpetual Graph Exploration problem (CPGE). That problem requires each robot to perpetually visit all the vertices of the graph, in such a way that no vertex hosts more than one robot at a time, and each edge is traversed by at most one robot at a time. The paper states an tipper bound k oil the number of robots that can be placed in the graph while keeping CPGE solvability. To make the impossibility result as strong as possible (no more than k robots can be initially placed in the graph), this upper bound is established under a strong assumption, namely, there is an omniscient daemon that is able to coordinate the robots movements at each round of the synchronous system. Interestingly, this upper bound is related to the topology of the graph. More precisely, the paper associates with each graph a labeled tree that captures the paths that have to be traversed by a single robot at a time (as if they were a simple edge). The length of the longest of these labeled paths reveals to be the key parameter to determine the upper bound k oil the number of robots. (C) 2008 Elsevier B.V. All rights reserved