Abstract
This article briefly discusses some of the changes that are taking place in systems architectures and the Internet due to the growth of distributed systems and SOA (service-oriented architectures). These changes include innovations such as Grid Computing, and may lead to a new global electronic web-service infrastructure.
The nature of the Internet is evolving: up until now, its primary role has been to pass web pages between servers and browsers for a human audience, but SOA and Grid computing are changing this. Web pages will soon form only the tip of the iceberg of Internet traffic.
SOA extends the well-known ‘client-server’ model. Software functionality is organised into modular ‘web-services’, and is used by a range of client components via a network or the Internet. These services might process data, do calculations or perform a wide range of other tasks. It then becomes progressively easier to engineer solutions that leverage off, and re-use, existing components and services.
We can now expect the emergence of a new dynamic global electronic web-service matrix, a new paradigm that will fundamentally change systems engineering. The impact of this may be as significant and unpredictable as that of the Internet to-date.
One approach, known as ‘grid computing’, involves using a large numbers of small computers working together in a group. This is becoming increasingly attractive, with the ready availability of powerful, low cost PCs.
Examples range from the large, co-located, ‘server farms’ used by search engines such as Google, to the climate modelling project, in which the spare processing power of many home PCs connected to the Internet are used to run calculations.
A design favoured by the author uses ‘grid-service-spaces’: to call a service, we simply insert a request into ‘grid-space’, from where it is taken and processed. In the author’s architecture, the elements that perform work are ‘mobile agents’, which move between the available hardware platforms, through grid-space, as necessary.
Examples on the RCRT web site include:
Simulating the breaking of the Enigma code.
Fast risk modelling and analysis algorithms.
A ‘GAVONTS’ genetic algorithm that can create an engine controller which can successfully land a spacecraft on a planetary surface (in a simulation).
Andrew Gray has PhD in astrophysics from Cambridge University. Before founding RCRT, he has been the head of Group Risk-Adjusted Portfolio Analysis, and the head of Group Risk Management Technology for a UK banking group. Andrew has extensive experience in quantitative modelling, systems design, software development and risk analysis. He is also available to undertake development and consultancy work.