National GRID Initiative

GRID computing (which is also known as a Computational Grid) is a distributed computing technology created in 1994-95 in the National Argonne Laboratory in the USA by Ian Foster, Steven Tuecke and Carl Kesselman, who are considered as the “fathers” of GRID computing.

The idea behind GRID computing is to replicate computational processing using the philosophy and principles of how the World Wide Web (WWW) works in order to make information available at a worldwide level. In actual fact, the WWW has made it possible for an enormous amount of information to be available in distributed form and via thousands of regular computers, which would be impossible to do with a small number of computers, even if they were extremely powerful. Similarly, GRID Computing manages to make high level computational processing available by distributing processing tasks between different computers in a coordinated and efficient manner.

GRID computing has created a revolution in how work is carried out in certain areas of science and technology. At a time when academic work in any area of knowledge produces ever more information, for example with the decoding of the human genome, the search for a cure for contagious, immunological or carcinogenic diseases, or the study of the behaviour of matter and energy in experiments involving particle accelerators, where the data from these are analysed and shared by researchers located around the world, GRID computing has been able to meet the enormous demands placed on computational processing and storage which the processing of large quantities of data require.

Furthermore, with large scale computing simulation systems, such as those used in meteorology, oceanography, genomics and proteomics, GRID Computing enables distributed computing resources to be linked up to solve problems which require a high level of computing performance.

In December 1994 the CERN Council approved the construction of a new machine which would enable a significant jump in high energy physics: The LHC – Large Hadron Collider. Construction work on the LHC started in 1998 with initial operation scheduled for the first half of 2008, involving experiments which had the aim of clarifying questions such as: What gives mass to matter? What is the invisible matter which forms 96% of the Universe made of? Why does nature prefer matter to anti-matter? How did matter evolve in the first moments after the Universe came into being?

The enormous quantity of computational processing necessary to analyse the experimental data which would be produced by the LHC, estimated at 15 Petabytes - 15 million Gigabytes - a year, for 15 years, soon led the CERN scientists to worry about how they could guarantee the necessary computing capacity to do this. GRID Computing seemed to be the appropriate technology.

For this reason the European DataGrid Project, financed by the European Research Framework Programme and coordinated by CERN, was started in 2001, which was later replaced by the EGEE project – Enabling Grids for E-science in Europe.

GRID Computation also provides a fresh vision for the creation of new products and services, affecting the way companies conduct their business.

The National GRID Initiative (INGRID) was planned and is monitored and partially financed by Knowledge Society Agency (UMIC), and R&D projects are carried out through the financial support of the Science and Technology Foundation (FCT), following a public tender for projects and independent international evaluation.

The main aims of INGRID are to:

• Strengthen national competencies and abilities in GRID computing in view of its special strategic importance.
• Strengthen national competencies and abilities and enable Portugal to be integrated within the international GRID Computing network.
• Improve conditions for scientific activities and for applications which have a social and economic interest and which include complex computing and/or high levels of data.
• Strengthen the multidisciplinarity and cooperation between research communities and users of high performance computer facilities.
• Strengthen the means for companies to locate scientific institutions and human resources in Portugal with knowledge and experience of GRID Computing.

15 R&D projects are currently underway representing total FCT funding of around 1.7 million Euros, which were selected following an open public tender in November 2006 which resulted in 37 bids.

These projects involve Grid Computing applications in areas which range from data simulation and analysis of high energy Physics such as those produced by the LHC and plasma physics and nuclear fusion, to forecasting the evolution of the maritime coastline, simulating forest fires, mapping atmospheric pollution, protein structure simulation, medical applications storage, and brain imaging.

The launch of the National GRID Initiative was accompanied by the publication of a guideline document on the initiative which took place in a public session held in the Pavillion of Knowledge which included various presentations on different European GRID computing projects, as well as a debate involving researchers and companies along with the signing of a protocol between UMIC and IBM Portugal concerning this company’s willingness to offer support software to distributed computing for use by the scientific community.

The 1st tender held under the scope of the National GRID Initiative (text in Portuguese) took place on 11 November 2006 at Minho University at the meeting INGRID '06 Journey (text in Portuguese).

At the Luso-Spanish Summit in November 2006 it was decided to join together the GRID Computing infrastructure of the two countries into one organised network, IBERGRID – The Iberian Computing Grid Network – to increase computational capacity and strengthen cooperation between the institutions and researchers from the two countries active in this field. As one of the instruments for intensifying collaboration between Spanish and Portuguese institutions and researchers, the decision was taken to hold a series of annual conferences, alternating from one country to the other. The 1st Conference held under the scope of IBERGRID - The Iberian Computing Grid Network - was held from 14 – 16 May 2007 in Santiago de Compostela.

Portugal participates in the EGEE – Enabling Grids for E-sciencE in Europe Project through LIP – the Experimental Particle Physics and Instrumentation Laboratory (Lisbon and Coimbra), of the Universities of Porto and Minho, the Centre for Plasma Physics at IST (Instituto Superior Técnico), and IEETA – the Institute of Electronics and Telematics Engineering of the University of Aveiro, and the Lusíada University (Famalicão).

THE EGEE project brings together scientists and engineers from more than 240 scientific institutions in 45 countries to provide a permanent GRID Computing structure which consists of 41 000 CPUs and 5 Petabytes disc space (5 million Gigabytes), involving the simultaneous functioning of 100 000 computational processes. The project initially focused on high energy Physics processes and life sciences, and now includes applications from other areas, such as Geology, computational Chemistry and image processing.

GRID Computing Jobs Carried Out for the EGEE Project - Enabling Grids for E-sciencE in Europe at Portuguese Sites
January 2006 - July 2008, Number of Jobs.

Alternative access:
press the image below to enlarge this

Source: EGEE Accounting Portal.

Using financing from Knowledge Society Agency (UMIC) and the Operational Programme for the Knowledge Society (POSC), it was decided to establish a main node for the GRID infrastructure at the Foundation for National Scientific Computation (FCCN),  which involved setting up a large 400 m2 GRID data centre, and increasing the Portuguese Grid infrastructure to around 650 CPUs by June 2008 and more than 1,200 CPUs by the end of 2008. This provides Portugal with a significant European-level capacity.

The National GRID Initiative also envisaged strengthening national and international connectivity with a significant increase in the bandwidth of the national Science and Education Network, the Science, Technology and Society Network (RCTS), which is managed by FCCN and financed by UMIC.

This increase in connectivity has occurred in several stages: the installation of fibre optic cable from FCCN in Lisbon to Braga (in 2005) and to Valença (in 2007) enabling connections of up to 10 Gigabits per second (Gbps) between research institutions and universities located within this coastal area and allowing connection to the Geant2 European teaching and research network through Spain, the connection of RCTS to the Geant2 network at 2.5 Gbps at the end of 2005, the extension of the FCCN fibre optic cable to include Lisbon-Setúbal-Évora-Portalegre-Fronteira do Caia which took place in the following six months, also with financing from UMIC and POSC, and ensuring fibre optic redundant ring connection to the Geant2 network through Spain, and the connection of RCTS to the Geant2 European Network at 10 Gbps by June 2008.

Of note at the level of international cooperation was the decision taken at the 2006 Luso-Spanish Summit to integrate the GRID computing infrastructures of the two countries into one extended network – the Iberian Computing Grid Network (IBERGRID), and the participation of Portugal, through LIP, in the main European GRID Computing projects using European Union (EU) funding, specifically through the aforementioned European DataGrid (2001-2004) and EGEE (stages I, II e III, 2004-2010) projects, and also through the CrossGrid (2002-2005), EELA (2006-2007), and Int.Eu.Grid (2006-2008) projects. It should be noted that EELA is an EU cooperative Project with Latin America (Argentina, Brazil, Chile, Cuba, Mexico, and Venezuela).

Also of note is the involvement in the LCG – LHC Computing Grid project (2003-2023) of CERN which forms the largest GRID Computing system in the world, where  LIP runs one of the Tier2 world centres, of which there are around 60 (the LHC also has around 11 Tier1 centres and the Tier0 centre is operated directly by CERN).

The union of Portugal and Spain within IBERGRID has also enabled them to form a network together known by the EGEE project as the South-Western European Federation, which currently has more than 1,500 CPUs and which is now playing a significant role in the GRID infrastructure of Europe.

Further examples of GRID Computing applications come from those developed within the CrossGrid project, the central management systems forming its infrastructure being located in Portugal under the operational direction of LIP - the Experimental Particle Physics and Instrumentation Laboratory (Lisbon and Coimbra). The applications within the CrossGrid project include control, forecasting and simulation of floods in river basins, atmospheric pollution modelling and other meteorological support services, as well as a vascular surgery support application. With this last application, using a patient’s CAT or Magnetic Resonance images, a surgeon can simulate the effect of a by-pass and visualise blood flow, deciding on the best solution for the intervention after carrying out successive simulations. This application enables the intervention time to be cut to a few hours between the image being taken and the optimised surgical intervention using prior simulation.

Another example, within the scope of the EGEE – Enabling Grids for E-sciencE in Europe project, was the computational simulation of components for possible treatment for Bird Flu which involved the analysis of 300,000 possible drugs against the H5N1 bird flu virus. Using GRID computing to help identify the most promising connections for biological tests may accelerate the development process of remedies against the virus. The simulations took place within the EGEE project GRID infrastructure for a month, while they would have taken around a century on a single computer. A relational database was created and saved involving more than 60,000 files with a total volume of information of 600 Gigabytes.

Another interesting example of an EGEE project was the setting up of one of the largest multimedia metadata collections in the world. In just 16 weeks, 37 million images were processed, which generated around 112 million text and image objects – almost 5 Terabytes (5 thousand Gigabytes) of data – containing altogether more than 150 million features extracted through automatic computational analysis of the images. This number is equivalent to the processing of 300,000 images per day. This type of technique enables the expansion of conventional search engines so as to be able to carry out searches on images and audiovisual content, even when these are not linked to text. It also contributes towards an automatic text description of images for the blind or for those who for whatever reason cannot see the corresponding images. The Portuguese LIP participated in the project, making GRID computing resources available to carry out the setting up of descriptive metadata for images from their computational analysis.

A two year project funded by the EU was started in 2007 to prepare for the setting up and modelling of the organisation of the EGI – European Grid Initiative. This project involves the national GRID initiatives of 38 countries, including the 27 EU countries as well as Belarus, Croatia, Israel, Moldova, Montenegro, Norway, Russia, Serbia, Switzerland, Turkey and the Ukraine. Portugal is represented by UMIC and by LIP, the senior figures of which form part of the GRID European Initiatives Policy Council, to which Professor Gaspar Barreira, the Director of LIP, was elected President.

In less than two years, Portugal has put itself on the European GRID Computing map and ensured the means to participate through a position of influence in decisions concerning the future of the European GRID initiative and the gigantic GRID Computing system which is being set up.