This is a continuation of a previous post and focuses on one of six tenets identified in the previous post. In this post I will focus on service orientation.
In today’s environment information can not be processed fast enough. After a decade of throwing bigger equipment at processing challenges, the community has still not come close to meeting the sixteen times real-time performance requirements established by the operational community. To reach threshold and keep up with the ever-increasing demands on complex information processing and analysis, a new approach is required. Distributed and parallel processing techniques offer a viable and scalable solution.
Over the last decade, organizations such as, SETI (the Search for Extraterrestrial Intelligence) and RSA data security have successfully used distributed computing to analyze data. This is accomplished by harnessing the idle CPU cycles of volunteer or subscriber computers/computing appliances distributed across the Internet which can process in excess of 100 million instructions per second. In the last 20 years, parallel processing, concurrent processing, and optimized distributed computing have realized vast savings in computational time for applications of wide variety. The problem in deploying parallel processing on the Web has been software limitations and reliance on unique hardware requirements leveraged by prevailing parallel programming techniques. For example, operating systems, such as Windows NT supported multiple processors, but desktop PC applications were slow to fully exploit its internal multi-threading capability.
Today, hardware innovations have led to high-speed switched interconnections within a wide range of computing devices. These interconnections have made distributed-memory massively parallel processing (MPP) appear to programmers like shared memory symmetric multiprocessing (SMP) machines. These hardware advances in addition to advances in Rapid Application Development (RAD) environments, have made the realization of this capability much easier. Forward looking approaches today, seek to leverage the rapid advances in processor design, Internet connectivity, and the implementation of distributed computing embodied in such languages as the .NET Framework and Java. In the systems of the Future, processing agents installed on available computing devices will allow processing of data such as acoustic signals and electronic emissions, more rapidly as available network assets increase. When combined with presence and discovery approaches discussed later in section, processing clients can automatically subscribe to process pools and provide more adaptive processing support.
Tags: distributed processing