Currently, most computational grids (systems allowing transparent sharing of processing resources across organizational boundaries) are assembled using human negotiation. This procedure does not scale well, and is too inflexible to allow for large open grids. Peer-to-peer grids present an alternative way to build grid infrastructures with a large number of sites. However, to actually assemble a large grid, peers must have an incentive to collaborate by providing resources to the system. In this paper we propose an incentive mechanism called the Network of Favors, which is designed to make it in the best interest of each participating peer to make as much as possible of its spare processing resources available to the system. We show through analysis, simulations, and experiments with an implementation that the Network of Favors promotes collaboration in a simple and robust fashion. We also describe OurGrid, a peer-to-peer grid based on the Network of Favors that is in production since December 2004.

Abstract: Large distributed systems challenge traditional schedulers, as it is often hard to determine a priori how long each task will take to complete on each resource, information that is input for such schedulers. Task replication has been applied in a variety of scenarios as a way to circumvent this problem. Task replication consists of dispatching multiple replicas of a task and using the result from the first replica to finish. Replication schedulers (i.e. schedulers that employ task replication) are able to achieve good performance even in the absence of information on tasks and resources. They are also of smaller complexity than traditional schedulers, making them better suitable for large distributed systems. On the other hand, replication schedulers waste cycles with the replicas that are not the first to finish. Moreover, this extra consumption of resources raises severe concerns about the system-wide performance of a distributed system with multiple, competing replication schedulers. This paper presents a comprehensive study of task replication, comparing replication schedulers against traditional information-based schedulers, and establishing their efficacy (the performance delivered to the application), efficiency (the amount of resources wasted), and emergent behavior (the system-wide behavior of a system with multiple replication schedulers). We also introduce a simple access control strategy that can be implemented locally by each resource and greatly improves overall performance of a system on which multiple replication schedulers compete for resources.

Several systems adopting Peer-to-Peer (P2P) solutions for resource discovery in Grids have recently been proposed. This report looks at a P2P resource discovery framework aiming to manage various Grid resources and complex queries. Following the discussion on characteristics of Grid resources and related query requirements, a DHT-based framework leveraging different P2P resource discovery techniques is proposed. The goal of the proposed framework is two-fold: to address discovery of multiple resources, and to support discovery of dynamic resources and arbitrary queries in Grids. 1

Abstract. As Grids enlarge their boundaries and users, some of their functions should be decentralized to avoid bottlenecks and guarantee scalability. A way to provide Grid scalability is to adopt Peer-to-Peer (P2P) models to implement non hierarchical decentralized Grid services and systems. A core Grid functionality that can be effectively redesigned using the P2P approach is resource discovery. This paper proposes a P2P resource discovery architecture aiming to manage various Grid resources and complex queries. Its goal is two-fold: to address discovery of multiple resources, and to support discovery of dynamic resources and arbitrary queries in Grids. The architecture includes a scalable technique for locating dynamic resources in large-scale Grids. Simulation results are provided to demonstrate the efficiency of the proposed technique. 1

Dynamic and heterogeneous characteristics of large-scale Grids make the fun-damental problem of resource discovery a great challenge. This thesis presents a self-organized grouping (SOG) infrastructure that achieves efficient Grid resource discovery by forming and maintaining autonomous resource groups. Each group dy-namically aggregates a set of resources that are similar to each other in some pre-specified resource characteristic. The SOG method takes advantage of the strengths of both centralized and decentralized approaches that were previously developed for Grid/P2P resource discovery. The design of the SOG method minimizes the overhead incurred in forming and maintaining groups and maximizes resource discovery perfor-mance. The way SOG method handles resource discovery queries is metaphorically similar to searching for a word in an English dictionary by identifying its alphabetical groups at the first place and then performing a lexical search within the group. The algorithms implemented in SOG method are illustrated with details. This thesis also illustrates a generalized approach using a space-filling curve

Dedicated to my wife and my parents

stream processing, overlay, quality-of-service

apport de recherche

Within computational grids, some services (software components, linear algebra libraries, etc.) are made available by some servers to some clients. In spite of the growing popularity of such grids, the service discovery, although e cient in many cases, does not reach several requirements. Among them, the exibility of the discovery and its e ciency on widearea dynamic platforms are two major issues. Therefore, it becomes crucial to propose new tools coping with such platforms. Emerging peerto-peer technologies provide algorithms allowing the distribution and the retrieval of data items while addressing the dynamicity of the underlying network. We study in this paper the service discovery in a pure peer-to-peer environment. We describe a new trie-based approach for the service discovery that supports range queries and automatic completion of partial search strings, while providing fault-tolerance, and partially taking into account the topology of the underlying network. We validate this approach both by analysis and simulation. Traditional metrics considered in peer-topeer systems exhibits interesting complexities within our architecture. The analysis ' results are con rmed by some simulation experiments run using several grid's data sets. Keywords: Service discovery, computational grids, peer-to-peer, pre x trees

The information service is an important component of any Grid middleware as it helps users to find resources suitable for the execution of their applications. Traditional approaches based on data bases that have been designed for Grids of moderate size do not fit the requirements of wide-area large scale Grids composed of dozens of thousands highly dynamic nodes. In this paper, we propose a fully distributed information service relying on an unstructured overlay network. The core of this information service is the RW-OGS cached-based optimized random walk protocol. We describe experimental results showing that RW-OGS outperforms state of the art protocols in terms of bandwidth consumption and resource discovery latency. Moreover, RW-OGS is optimized to increase the probability to discover nodes with free resources and to enable fast discovery of rare resources. The RW-OGS protocol has been implemented as part of the Vigne middleware aiming at easing the use of large scale desktop Grids for executing a wide range of distributed applications. 1.