Summarizing CCGrid 2007 From: http://www.gridtoday.com/grid/1677376.html By Bruno Schulze, LNCC (Brazil); and Omer Rana, Cardiff University For the first time, the Cluster Computing and Grid (CCGrid 2007) conference was held in Latin America, specifically in Rio de Janeiro, Brazil. The event, which took place May 14-17, included strong participation from the Latin American grid community. Overall, 330 people attended the event, a number greater than any other CCGrid event in its six-year history. This is a testimony to the fact that the grid and cluster computing communities continue to grow, and many early projects that were started in 2001 (when the conference series began) are now maturing and producing significant results. The CCGrid07 main conference track received 218 submissions from 38 different countries, of which 72 were selected for presentation. Traditionally, the CCGrid series has been held in countries with recognized research centers, such as Australia, Germany, Japan, the United States, the United Kingdom and Singapore, so the hosting of CCGrid in Brazil is a definite recognition of Brazil as one of the world leaders in R&D in the field of cluster and grid computing. In many ways, this is thanks to the extensive work taking place at Brazilian universities and research laboratories. The CCGrid 2007 conference was organized by the National Laboratory for Scientific Computing (LNCC, Brazil), the University of Melbourne (Australia) and the Federal University of Rio Grande do Sul (UFRGS, Brazil). Grid computing started as a generalization of cluster computing, promising to deliver unprecedented levels of parallelism to high-performance applications by crossing administrative boundaries. Subsequently, this vision evolved to support on-demand access and composition of any computational service, provided by multiple independent sources. Under this new vision, clusters gained renewed importance as the "super-servers" of the emerging grid infrastructure. Meanwhile, the use of computational and data resources in high-performance applications, undertaken over grid infrastructure, is now starting to become a reality. Today, we face the huge challenge of making on-demand access to any computational service, the "computing as a service" vision, a widespread reality. The CCGrid symposia have been part of this journey, bringing together researchers and practitioners and enabling them to share their insight, results, and experience in the multi-faceted areas of grid and cluster computing. CCGrid 2007 areas of interest included: grid economies and service architectures; grid architectures and systems; utility computing models for clusters and grids; middleware for clusters and grids; programming models, tools and environments; resource management; performance evaluation and modeling; peer-to-peer systems; grid-based problem solving environments; grid trust and security; service composition and orchestration; community networks; community and collaborative computing networks; scheduling and load balancing; scientific, engineering and commercial applications; and parallel and wide-area file systems, among others. CCGrid 2007 keynote presentations included: "Grids Sandwiched by Web 2.0 and Multicore" by Geoffrey Fox, Indiana University; "Towards an International ‘Computer Science Grid’" by Dr. Franck Cappello, INRIA (France); and "Scale-up and Scale-out: Evolution and Trends in Parallel Processing" by José E. Moreira, IBM Thomas J. Watson Research Center. Fox outlined the need for the grid community to address the challenges of multi-core processor architectures, and outlined that with such computational power available, one could think of a "grid on a chip." He went on to discuss the types of programming models that would be needed to support such a vision. Fox also outlined the emerging interest in Web 2.0 technologies (mainly centered on social networking themes), and illustrated the large number of users that are now involved in projects making use of such technologies. He emphasized the need for grid computing researchers to better understand the demands of this community, and also suggested that current -- and excessive -- focus on Web services standards may be a potential bottleneck for future grid computing growth. Cappello outlined the work being undertaken in the French Grid5000 project, which involves aggregating computation capacity across France to provide a unique "computer science grid". As a director of this project, his vision was unique from that of the National Grid Service in the United Kingdom and the TeraGrid in the United States. His focus in Grid5000 was to allow computer scientists to evaluate new algorithms on distributed resources and have exclusive access to such resources by pre-booking/reserving them. The OAR tool was developed within the Grid5000 project to achieve this. Using this approach, a researcher could re-boot machines on the Grid5000 (when they had reserved time on these) and install custom software on these machines for their own work. The Grid5000 also supports a Unix-style file sharing mechanism, whereby users see a distributed file space and can use this to transfer data transparently between machines. Cappello also outlined the novel work in the Dutch DAS-3 project, which involves computational clusters in different parts of the Netherlands connected over an optical backbone network (via programmable optical switches). This is one of the first grid infrastructures that allows programmers to directly program the underlying network switches for their own applications. Recent efforts to connect Grid5000, DAS-3 and the Japanese NAREGI grid also were briefly described. Grid5000 continues to be a novel and highly ambitious project, and one that needs to be considered more closely by grid communities in other parts of the world. In parallel with the main conference track, there were several workshops on emerging topics, as well as the first IEEE TCSC Doctoral Symposium. There were three workshops on ongoing editions: Agent-Based Grid Computing (AGC07); Biomedical Computations on the Grid (Biogrid07); and Global and Peer-2-Peer Computing (GP2P07). New workshops also were introduced on exciting topics and regional context: Context-Awareness and Mobility in Grid Computing (WCAMG07) and Programming Models for Grid Computing (PMGC07). In the Latin American context there was a first edition of the Latin American Grid Workshop (LAgrid07) with invited speeches on "The Current Situation and Perspectives for e-Science in Latin America" by Michael Stanton from RNP (Brazil), and "Towards Truly Ubiquitous Cyberinfrastructure" by Jim Myers from NCSA, and also selected papers. Additionally, we had an industry track with presentations on: "Google Infrastructure for Massively Parallel Processing" by Walfredo Cirne from Google; "Promoting Cooperation in BitTorrent Communities" by Miranda Mowbray from HP Labs Bristol; and "Opportunity and Challenges in e-Science" by Fabrizio Gagliardi and Carlos Hulot from Microsoft Corp. The MapReduce function now being supported by Google provides an interesting way in which industry is adopting functional programming approaches to undertake specific parallel functions. Cirne outlined how the MapReduce function was being parallelized across machines at Google, and how this could be used as a basis to introduce additional grid computing concepts within the company. Mowbray discussed the emerging interest from HP Labs in social networking and outlined how these approaches were being used to investigate legal concerns in file sharing within a community of users. A variety of "sharing patterns" within BitTorrent were used to illustrate some of the ideas. Gagliardi, who formerly led the European Datagrid project, discussed the recently established "technical computing" initiative at Microsoft, as well as the emerging interest from Microsoft to provide additional capability within Microsoft software to support clustering applications. CCGrid 2007 also offered five tutorials on the topics of: autonomic grid computing; MOSIX cluster and grid management system; open source middleware for the grids; ObjectWeb ProActive; designing clusters and grid computing systems with InfiniBand and iWARP; and Gridbus Toolkit for building and managing utility grids for powering e-science and e-business applications. The "Best Sustained Technical Contribution Award" was given to Professor Ken Kennedy and his team at Rice University. He passed away recently and he had at least one paper in each CCGrid conference since its inception in 2001. Some authors also will be asked to contribute extended versions of papers to a special issue of the Concurrency and Computation: Practice and Experience journal. At the closing session, we had impressions and remarks about the conference given by Geoffrey Fox, Andrew Grimshaw (University of Virginia) and Joerg Schneider (Technische Universität Berlin). This summary of CCGrid07 cannot fully cover the variety of work presented at the conference. An attempt is made to present some key themes at the conference, and to highlight some new and emerging areas of interest to the community. Interested readers should seek out the proceedings of CCGrid 2007 for additional information. The next CCGrid conference is scheduled to take place in May in Lyon, France.