4th EGEE User Forum/OGF 25 and OGF Europe's 2nd International Event

Session

Grid Research

2 Mar 2009, 17:30

Le Ciminiere, Catania, Sicily, Italy

Description

At EGEE-08, EGEE director raised three questions: "How can we reduce the effort required to operate this expanding infrastructure?", "How do we match the expectations of the growing user communities?" and "Will we have a European Infrastructure by the end of 2009?". Some grid research themes are extremely relevant to these questions: fault diagnosis, detection, and tolerance are obviously related to operations; Quality of service (QoS) oriented resource allocation is critical for user satisfaction; both are required to ensure organised resource sharing and avoid fragmentation based on national or community resource ownership.

This session will provide insights on new and exciting work contributing to the above-mentioned goals. As a grid research session, the talks will focus on methods and proof of concepts more than finalised products. Nonetheless, realistic hypothesis and quantitative evaluations are present in all the contributions, thus connecting these works with the day-to-day problems. Benchmarking, which is the basis of a quantitative approach of middleware design, is exemplified. The potential of a Machine Learning (a sub-discipline of Artificial Intelligence) is demonstrated in the areas of, on one hand QoS, and on the other hand, fault diagnosis. Finally, QoS is also addressed through the SLA framework, in a work proposing a new paradigm for resource allocation, which is a hot topic in both grid and cloud research.

Presentation materials

slides

IntroSessGridResearch.pdf

IntroSessGridResearch.ppt

5. Design of an Expert System for Enhancing Grid Fault Detection based on Grid Monitoring Data

Ms Gerhild Maier (Johannes Kepler Universität Linz)

02/03/2009, 17:30

Planned or on-going scientific work using the grid

Oral

Grid computing is associated with a complex, large scale, heterogeneous and distributed environment. The combination of different Grid infrastructures, middleware implementations, and job submission tools into one reliable production system is a challenging task. Given the impracticability to provide an absolutely fail-safe system, focusing on strong error reporting and handling is a crucial...

16. Performance Analysis of Existing and Emerging Computing Interfaces for Grid Computing

Mr Laurence Field (CERN)

02/03/2009, 17:50

Emerging Technologies within the EGEE infrastructure

Oral

The Grid interface to a batch system is one of the primary services in a Grid infrastructure. The performance of this interface is of importance for high though-put users and managers of large clusters. This paper defines a benchmarking methodology for evaluating the performance of these interfaces and which is used to benchmark a number of existing and emerging implementations.

88. Toward Responsive Grids through Multi Objective Reinforcement Learning

Julien Perez (LRI, CNRS and Université Paris-Sud)

02/03/2009, 18:10

Scientific results obtained using grid technology

Oral

EGEE has experimented in specialized software and configurations, such as priorities, Virtual Reservations and overlay task-management in order to provide differentiated Quality of Service (QoS) requested by an increasingly diverse user community. To combine differentiated QoS, fair-share and self-configuration under a unique production model, we propose a multi objective Reinforcement...

146. QoS management in Grids

Antonella Di Stefano (DIIT - Catania University)

02/03/2009, 18:30

Grid Services exploiting and extending gLite middleware

Oral

The grids are opening towards new application scenarios, embracing not only the e-science field but also business, financial and educational ones. This evolution requires that the grids have to be able to supply resources and services in a flexible manner, offering them on-demand to several different typologies of users, each one characterized by specific "Quality of Service (QoS)"...