Autor:
Zoltan, Cristina

Cargando...
Foto de perfil

E-mails conocidos

zoltan@cs.upc.edu

Fecha de nacimiento

Proyectos de investigación

Unidades organizativas

Puesto de trabajo

Apellidos

Zoltan

Nombre de pila

Cristina

Nombre

Nombres alternativos

Afiliaciones conocidas

Computer Science Department. Universitat Politècnica de Catalunya, Spain / Universidad Internacional de Ciencia y Tecnología, Panamá
Universitat Politècnica de Catalunya, Spain

Páginas web conocidas

Página completa del ítem
Notificar un error en este autor

Resultados de la búsqueda

Mostrando 1 - 3 de 3
  • Artículo
    Comparing MapReduce and Pipeline implementations for Counting Triangles
    Pasarella, Edelmira; Vidal, Maria Esther; Zoltan, Cristina. Actas de las XVI Jornadas de Programación y Lenguajes (PROLE 2016), 2016-09-02.
    A common method to define a parallel solution for a computational problem consists in finding a way to use the Divide & Conquer paradigm in order to have processors acting on its own data and scheduled in a parallel fashion. MapReduce is a programming model that follows this paradigm, and allows for the definition of efficient solutions by both decomposing a problem into steps on subsets of the input data and combining the results of each step to produce final results. Albeit used for the implementation of a wide variety of computational problems, MapReduce performance can be negatively affected whenever the replication factor grows or the size of the input is larger than the resources available at each processor. In this paper we show an alternative approach to implement the Divide & Conquer paradigm, named dynamic pipeline. The main features of pipeline are illustrated on a parallel implementation of the well-known problem of counting triangles in a graph. This problem is especially interesting either when the input graph does not fit in memory or is dynamically generated. To evaluate the properties of pipeline, a dynamic pipeline of processes and an ad-hoc version of MapReduce are implemented in the language Go, exploiting its ability to deal with channels and spawned processes. An empirical evaluation is conducted on graphs of different topologies, sizes, and densities. Observed results suggest that pipeline allows for the implementation of an efficient solution of the problem of counting triangles in a graph, particularly, in dense and large graphs, drastically reducing the execution time with respect to the MapReduce implementation.
  • Artículo
    The Dynamic Pipeline Paradigm
    Zoltan, Cristina; Pasarella, Edelmira; Araoz, Julian; Vidal, Maria Esther. Actas de las XIX Jornadas de Programación y Lenguajes (PROLE 2019), 2019-09-02.
    Nowadays, in the era of Big Data and Internet of Things, large volumes of data in motion are produced in heterogeneous formats, frequencies, densities, and quantities. In general, data is continuously produced by diverse devices and most of them must be processed at real-time. Indeed, this change of paradigm in the way in which data are produced forces us to rethink the way in which they should be processed even in the presence of parallel approaches. To process continuous data, data-driven frameworks are demanded; they are required to dynamically adapt execution schedulers, reconfigure computational structures, and adjust the use of resources according to the characteristics of the input data stream. In previous work, we introduced the Dynamic Pipeline as one of these computational structures and we experimentally showed its efficiency when it is used to solve the problem of counting triangle in a graph. In this work, our aim is to define the main components of the Dynamic Pipeline which is suitable to specify solutions to problems whose incoming data is heterogeneous data in motion. To be concrete, we define the Dynamic Pipeline Paradigm and, additionally, we show the suitability of our framework to specify the solution to different well-known problems.
  • Artículo
    Towards a Dynamic Pipeline Framework implemented in (parallel) Haskell
    Royo Sales, Juan Pablo; Pasarella, Edelmira; Zoltan, Cristina; Vidal, Maria Esther. Actas de las XX Jornadas de Programación y Lenguajes (PROLE 2021), 2021-09-22.
    Data streaming processing has given rise to new computation paradigms to provide effective and efficient data stream processing. The most important features of these new paradigms are the exploitation of parallelism, the capacity to adapt execution schedulers, reconfigure computational structures, adjust the use of resources according to the characteristics of the input stream and produce incremental results. The Dynamic Pipeline Paradigm (DPP) is a naturally functional approach to deal with stream processing. This fact encourages us to use a purely functional programming language for DPP. In this work, we tackle the problem of assessing the suitability of using (parallel) Haskell to implement a Dynamic Pipeline Framework (DPF). The justification of this choice is twofold. First, from a formal point of view, Haskell has solid theoretical foundations providing the possibility of manipulating computations as primary entities. From a practical perspective, it has a robust set of tools for writing multithreading and parallel computations with optimal performance. As a result, we present a dynamic pipeline to compute the weakly connected components of a graph (WCC) in Haskell (a.k.a. DP-Haskell). The behavior of DP-Haskell is empirically evaluated and compared with a solution provided by a Haskell library. The evaluation is assessed in three networks of different sizes and topology. Performance is measured in terms of the time of the first result, continuous generation of results, total time, and consumed memory. The results suggest that DP-Haskell, even naive, is competitive with the existing solution provided in the Haskell library. DP-Haskell exhibits a higher continuous behavior and can produce the first result faster. The observed results are encouraging and provide evidence of the benefits that Haskell's abstractions bring in implementing WCC and DPP. Built on them, we will develop a general and parametric DPF.