Catálogo de publicaciones - libros

The execution of scientific workflows in Grid environments imposes many challenges due to the dynamic nature of such environments and the characteristics of scientific applications. This work presents an algorithm that dynamically schedules tasks of workflows to Grid sites based on the performance of these sites when running previous jobs from the same workflow. The algorithm captures the dynamic characteristics of Grid environments without the need to probe the remote sites. We evaluated the algorithm running a workflow in the Open Science Grid using twelve sites. The results showed improvements up to 150% relative to other four usual scheduling strategies.

In this paper we describe the implementation of a service oriented environment that enables to couple a parallel application, which performs the 3D linear dynamic structural analysis of high-rise buildings, to a Grid Computing infrastructure. The Grid service, developed under Globus Toolkit 4, exposes the dynamic simulation as a service to the structural scientific community. It employs the GMarte middleware, a metascheduler that enables to perform the computationally intensive simulations on the distributed resources of a Grid-based infrastructure.

Parallel and Distributed Computing, Cluster and Grid Computing, Large Scale Simulations in All Areas of Engineering and Science.

Desktop grids are easy to install on large number of personal computers, which is a prerequisite for the spread of grid technology. Current desktop grids connect all PCs into a flat hierarchy, that is, all computers to a central server. SZTAKI Desktop Grid starts from a standalone desktop grid, as a building block. It is extended to include clusters displaying as single powerful PCs, while using their local resource management system. Such building blocks support overtaking additional tasks from other desktop grids, enabling the set-up of a hierarchy. Desktop grids with different owners thus can share resources, although only in a hierarchical structure. This brings desktop grids closer to other grid technologies where sharing resources by several users is the most important feature.

Analyzing the scalability behavior and the overheads of Open-MP applications is an important step in the development process of scientific software. Unfortunately, few tools are available that allow an exact quantification of Open-MP related overheads and scalability characteristics. We present a methodology in which we define four overhead categories that we can quantify exactly and describe a tool that implements this methodology. We evaluate our tool on the Open-MP version of the NAS parallel benchmarks.

This work presents an implementation of a parallel Fuzzy c-means cluster analysis tool, which implements both aspects of cluster investigation: the calculation of clusters’ centers with the degrees of membership of records to clusters, and the determination of the optimal number of clusters for the data, by using the PBM validity index to evaluate the quality of the partition.

The work’s main contributions are the implementation of the entire cluster’s analysis process, which is a new approach in literature, integrating to clusters calculation the finding of the best natural pattern present in data, and also, the parallel processing implementation of this tool, which enables this approach to be used with vary large volumes of data, a increasing need for data analysis in nowadays industries and business databases, making the cluster analysis a feasible tool to support specialist’s decision in all fields of knowledge.

The results presented in the paper show that this approach is scalable and brings processing time reduction as an benefit that parallel processing can bring to the matter of cluster analysis.

Unsupervised Classification, Fuzzy c-Means, Cluster and Grid Computing

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 () models to implement non hierarchical decentralized Grid services and systems. A core Grid functionality that can be effectively redesigned using the P2P approach is . 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.

In this paper we present JaceV, a multi-threaded Java based library designed to build asynchronous parallel iterative applications (with direct communications between computation nodes) and execute them in a volatile environment. We describe the components of the system and evaluate the performance of JaceV with the implementation and execution of an iterative application with volatile nodes.

In this paper, we present an approach to develop parallel applications based on aspect oriented programming. We propose a collection of aspects to implement group communication mechanisms on parallel applications. In our approach, parallelisation code is developed by composing the collection into the application core functionality. The approach requires fewer changes to sequential applications to parallelise the core functionality than current alternatives and yields more modular code. The paper presents the collection and shows how the aspects can be used to develop efficient parallel applications.

This paper proposes a new model to predict the process execution behavior on heterogeneous multicomputing environments. This model considers the process execution costs such as processing, hard disk acessing, message transmitting and memory allocation. A simulator of this model was developed which help to predict the execution behavior of processes on distributed environments under different scheduling techniques. Besides the simulator, it was developed a suite of benchmark tools in order to parameterize the proposed model with data collected from real environments. Experiments were conduced to evaluate the proposed model which used a parallel application executing on a heterogeneous system. The obtained results show the model ability to predict the actual system performance, providing an useful model for developing and evaluating techniques for scheduling and resource allocation over heterogeneous and distributed systems.

This paper evaluates the impact of the parallel I/O scheduling strategy on the performance of the file access in a parallel file system for clusters of commodity computers (Clusterfile). We argue that the parallel I/O scheduling strategy should be seen as a complement to other file access optimizations like striping over several I/O servers, non-contiguous I/O and collective I/O. Our study is based on three simple decentralized parallel I/O heuristics implemented inside Clusterfile. The measurements in a real environment show that the performance of parallel file access may vary with as much as 86% for writing and 804% for reading with the employed heuristic and with the schedule block granularity.