Case Studies

Pierre Sablonière Email Web Page

Ian Foster and Carl Kesselman outlined Grid concepts in 1999 in their book "The Grid Blueprint for a new Computing infrastructure". In a subsequent paper, "the Anatomy of the Grid", the following definition was proposed for Grid as "a Technology that provides coordinated resource sharing and problem solving in dynamic, multi-institutional virtual organizations". Looking further down in actual Grid flavours we have the following functional taxonomy: Compute Grid, Data Grid, Scavenging Grid, Aggregating Grid and Hybrid Grids. 
Examples of Compute Grids exist today such as the US TeraGrid, regrouping various Centers for Supercomputing Applications. A data grid example is the US National Digital Mammography Archive system, which provides access to huge and increasing data Décrypton in France or the smallpox project are examples of a virtual dynamic multi organization Grid with tens of thousands PCs federated to gather invaluable scientific data. 
In reality these examples are not trivial, as there are many issues to be considered when designing and setting a Grid. A Grid infrastructure has to address: Heterogeneity, integration (OS, protocols) - WANness (bandwidth, latency and disconnects) - Management (autonomy, policy-based) - Dynamic nature (unpredictability, transient) - Large data movement (flat files, RDBMS) - Security (privacy, credentials, authentication) - Large scale serial, batch, parallel processing - Implementing resilience (industrial strength) - Accounting and billing - Guaranteed Quality of Service (QoS). Today, the existing momentum set by the Internet success and I/T standard adoption thru Open sources and Linux operating system makes possible to envisage standardization of the Grid. This is being addressed by the Grid Global Forum (www.gridforum.org) with a paper "The physiology of the Grid" detailing an "Open Grid Services Architecture". The OGSA is based on a layered architecture and proposes the usage of Web Services standards, UDDI as directory and SOAP for service delivery. The open source Globus package (www.globus.org) is the privileged implementation. 
The Grid concepts are hot today. Grid is seen by CIOs as a game changer. For users, Grid provides access to a vast amount of resources. For business people, Grid will tap I/T power on demand like electricity is distributed today. Grid will not eliminate traditional I/T infrastructures it will gradually change the way we consider I/T solutions just like the Web changed forever the relationship we have with the Information Technology. Already today, Grid delivers with projects such as the Décrypton and the NMDA and others solution to very serious problems that could not have been solved by other means.

Brief Bio of Pierre Sablonière

Pierre Sablonière is the Technical Director of the IBM EMEA ATS Products & Solution Support Centre located in Montpellier, France. In his current responsibilities, he overlooks all critical projects and special initiatives occurring in the centre such as benchmarks and porting projects. He is technically heading the IBM Grid Innovation Centre opened in Montpellier late April 2002. Pierre Sablonière graduated from Ecole Polytechnique, Paris and holds a postgraduate degree in IT from the University of Montpellier. He is an IBM Senior Certified IT Specialist and a member of the IBM Technical Expert Council for the EMEA West region.

Marc Shapiro Email Web Page

Users often take replicas of their important data, for instance replicating mail folders and appointments from their PC to a handheld device. When information is updated the replicas become inconsistent. Our IceCube project aims to automate consistency in a general manner. To this goal, we designed what is in essence a scheduler that combines disconnected updates, from all applications, in a non-conflicting way, according to semantic constraints provided by applications. What started as blue-sky research has practical impact and has opened new theoretical avenues.

Brief Bio of Marc Shapiro

Dr. Shapiro graduated from ENSEEIHT, in Toulouse (France), in 1978, and received his Ph.D. from the Université Paul-Sabatier of Toulouse in 1980. After a post-doc at MIT, 1980--1982, he worked for the Centre Mondial Informatique et Ressources Humaines in Paris from 1982 to 1984. His collaboration with INRIA started in 1983; in 1985 he started the SOR group. He spent the 1993--1994 year on sabbatical at the Computer Science Department of Cornell University in Ithaca, NY (USA). He was the coordinator for the Esprit Long Term Research project PerDiS, a Persistent Distributed Store for Cooperative Engineering applications. He now leads the Cambridge Distributed Systems Group at Microsoft Research Ltd. in Cambridge (UK) since October 1998.

 
Page Updated on 11-10-2005

Copyright © Escola Superior de Tecnologia de Setúbal, Instituto Politécnico de Setúbal