Grid Computing

Grid Computing - Background

Many aspects of distributed computing have now become quite familiar and widely accepted with the growth of the Internet and related technologies. In addition, large-scale distributed and parallel computing projects such as SETI (the Search for Extra Terrestrial Intelligence), have also come to public prominence, in which a large number of computers around the world have been used together to search for 'signals' in observational data gathered by radio telescopes.

The author first became actively involved with these techniques - which later became known as Grid Computing - in 1999, when, as head of Group Risk (Management) Technology of a major UK bank, he identified, proposed and then (having acquired a large number of ex-desktop PCs) led a team which implemented a Distributed Service-Oriented Architecture using parallel-enabled analytics as a high performance/low cost solution to meet anticipated strategic requirements for risk-analysis and portfolio modelling.

Service-Oriented Architectures and 'The Grid'

Using many computers working together to provide one-another with a Portfolio of Services is the basis for what have now become known as: Service-Oriented Architecture (or SOA), and of Grid Computing.

Many of these techniques have now moved much closer to the main-stream. SOA Architectures are now widely accepted, and new technologies are available for implementing them. The use of processor farms and parallel computing techniques is now no-longer regarded as unreasonably creative or eccentric.

Such innovative solutions are now becoming even more attractive, with the ready availability of powerful, low-cost computers, arising from the demands of the PC market; with its continuing appetite for increased power and now unprecedented economies of scale. This, together with advances in the infrastructure technologies required to support robust distributed computing, have resulted in the emergence of the Grid as a practical, cost-effective, technological reality.

It is the opinion of the author, that as Grid Computing techniques are more broadly appreciated, they will become more and more widely adopted, and will ultimately lead to a fundamental change in application architectures, and how computers are used, especially in large-scale business environments.

For an article by the author on the evolution of these technologies, see the following :

• HTML format

However, the design and implementation of effective distributed, grid-based, parallel processing solutions is rarely straightforward, and properly taking advantage of grid-enabled technologies often requires a fundamental shift of thinking in the way that solutions are engineered.

The Possible Evolution of The Grid

In the genuinely distributed and diverse environment which SOA makes possible, many different software agents may function in an inter-dependant and logically inter-connected way, often spanning a range of different locations and hardware platforms.

As this evolution progresses, a new paradigm emerges. We can start to think of building software applications which run using an underlying electronic service infrastructure matrix which would be at their disposal. These services could provide applications with the ability to access data, perform calculations, and undertake a diverse range of other tasks.

The concept of The Grid then has several dimensions. From the hardware perspective, it might consist of a set of machines which could be homogeneous or heterogeneous in nature, physically co-located, or widely dispersed, be fixed or variable in number, and with a range of connection topologies mediated by dedicated network(s) and/or the Internet.

At one extreme this arrangement might consist of a small localised group of machines, working together as part of a particular application or application framework; while, at the other, there could be a large number of machines distributed over the internet, all working together in a more weakly-coupled scenario.

From the functional perspective we can think of the grid as providing a Service Infrastructure Matrix which could be used to deliver both data and functionality to applications, and to underpin the provision of new services.

Possible Future Developments ^*

Just as we have seen the growth of the Internet over the recent past, for the provision of web-pages to a human audience, there may well now be a similar growth in grid computing. Looking some way forward this means that there is the possibility of the emergence of a global web service-matrix which could have quite a wide-spread and unpredictable impact; and which might evolve in ways, and with benefits, that are not currently anticipated.

Since the next phase of the evolution of the Internet will include significant advances in communications between purely electronic peers, these developments could ultimately be more fundamental than those of the Internet to-date. New applications and pieces of software would then be able to build on, and leverage off, existing services increasingly effectively.

As a result of this, High Performance Computing techniques will become much more widely accessible, no longer being only available to those with access to supercomputers.

Computer systems should also be able to communicate much more easily and seamlessly than they do now. The emergence of ontologically-enabled service infrastructures will open up the possibility of systems being able to communicate electronically and process information based on its meaning in far more subtle ways than is currently possible.

It is also possible that the established relationship between Servers and Services will become increasingly outdated. The model which is favoured by the author is for grid-services to be provided by a population of Mobile Software Agents which would, where appropriate, be able to move around between the available hardware platforms, and replicate in a quasi-autonomous fashion. This is the way in which the author's Exciton Framework operates, which is based on our XML Space technology, and is designed to make these techniques a practical reality.

* This section represents the views of the author.