Вернуться в библиотеку
Перевод отредактировал Усачёв А.А.
На русском
The Global Grid Forum 5 - A Trip Report from Edinburgh
By Dr. Omer F. Rana, Cardiff University and Welsh e-Science Center
Printer Friendly
http://www.gridpartners.com/articleggf5.html
Overview
The Global Grid Forum (GGF) has continued to expand, at a truly alarming rate, since 2001 when it was formed. It is interesting to see the wide variety of interests and backgrounds at the GGF, and in particular, how these interests are forcefully expressed at working groups (WGs), research groups (RGs) and Birds-of-a-feather (BoF) sessions. Perhaps it is this diversity that has recently led to a lack of overall focus for the GGF (or atleast it appears to be this way to me) -- although I can also appreciate that it is important that a range of different opinions are aired (perhaps an important aspect lacking at related conferences and workshops). Perhaps, chairs of the various working areas should aim to provide a summary of common themes, and meet to discuss these after each GGF?
I attended various sessions at the GGF, and also participated in discussions during and after these sessions. The most important aspect was the discussions outside the sessions, allowing me to get a better understanding of the current state of Grid activities. It is clearly the case that the number of applications now utilising Grid technologies are increasing, and a larger number of researchers are providing "Grid-enabled" versions of their tools. Some of the new tools still need to be widely deployed and tested -- although initial enthusiasm is certainly promising! Industry interest has also grown considerably over the last year -- although the exact business models that are needed are not always openly mentioned or discussed (perhaps, this is a secret each company wants to maintain for themselves -- although a more sceptical analysis is also possible). The number of Grid projects in other countries is also interesting to see - ranging from the Asia-Pacific region, China, India, various European nations, and the Middle East. The only common denominator appears to be Globus and a lot of enthusiasm -- although these are certainly good starting points for useful collaborations in the future.
Service Management Frameworks -- Research Group
This was the first session for the SMF research group (previously called the Jini Working Group) -- and 3 sessions were devoted to activities of the group. There were a total of 9 presentations, and over 60 people attended the sessions. The group was created to move away from a single system (single vendor) -- to enable aspects available in systems such as Jini, Salutation, UPnP, JXTA, .NET, CORBA etc to be understood in a better way -- by analysing common themes and requirements which these systems try to address. Although Jini activities are still being encouraged, participants utilising other approaches are also now allowed to present and discuss their work. The group has also decided to take a longer term perspective on "services" -- and in particular how various services from different systems may be more easily combined. The group now aims to develop better collaboration with other groups at the GGF -- and to better understand how services may be created/destroyed, grouped, discovered, composed and executed -- using a range of different technologies and paradigms. The group does not have an overall consensus of particular features that should be supported within a management framework -- although a survey document is under preparation, and to be presented by GGF7 in March 2003. We hope this will provide a useful discussion "initiator" within the GGF. The document will be circulated on the group Web site to also receive comments from a wider community.
Grid Computing Environments
Sessions of the GCE group were attended by a large number of people -- and the interest in activities of this group continues to increase. A number of interesting presentations were given, although there was no overall consensus on the way forward. The topics discussed ranged from workflow systems, common component models, means of supporting and managing component execution, and the emerging interest in relating components with services.
Invited Talks
I attended three invited talks (plenaries), and found these very useful and interesting to identify future directions for Grids research. Jeff Nick and Ian Foster outlined the next steps for the OGSA, and the importance of integrating and specifying interfaces to services in a common format (such as WSDL). The important of UDDI as a registry mechanism was, although, played down. Work undertaken in other working groups was highlighted, particularly where there is likely to be a direct impact on OGSA developments. It is interesting that service discovery is seen as a key theme within OGSA, although there is no mention of how this is to be achieved.
Rick Stevens from Argonne provided a thought provoking and informative talk about bio-informatics, and the next series of activities that are to be undertaken at Argonne in this area. The talk covered a number of aspects of integrating computing and biology -- and the focus was not particularly on what one would have expected! (i.e. the analysis of genomic sequences -- such as structure alignment, microarray analysis/expression profiling etc). The talk covered aspects of biological ideas of the cell (and its various structures such as the mitochondria, RNA transport etc) -- as nano-factories, that could provide useful suggestions for future nano-technologies. The importance of the Grid was identified to be two folds: (i) to support the construction of these new ``nano-factories'' and support the associated simulations of these on a large scale, and (ii) to undertake computing with these new elements -- to fuel the emerging area of embedded and ubiquitous computing. Another aspect highlighted was the understanding of metabolic pathways to suggest data transfer in a more efficient manner over the Grid.
Fran Berman from SDSC outlined the requirements for a future Grid -- and related this to existing projects being undertaken as SDSC and as part of the NPACI alliance. The emphasis ranged from integration of specialist devices and wireless networks into the Grid (based on existing projects being undertaken at SDSC and the San Diego area), to the need to have reliability and dependability of infrastructure supporting the Grid. Fran Berman outlined various experiments that were undertaken by Casanova et al. to put together a host of different platforms -- with different architectures, operating systems, environments etc - and demonstrated the importance of bringing out real working examples to show to the community in the Grid area. She also suggested the importance of supporting heterogeneous computing environments, and to enable a number of vendors to work together. Also highlighted in the talk was the importance of better collaboration between the computer science and application science communities, and to allow these collaborations to feed into the software that develops to support Grid applications.
Acknowledgment
I would like to thank the National E-Science Centre in Edinburgh, UK for providing the travel grant.