"Exerted Enterprise Computing: from Protocol-oriented Networking to Exertion-oriented Networking"
From the very beginning of networked computing, the desire has existed to develop protocols and methods that facilitate the ability of people and automatic processes across different computers to share resources and information across different computing nodes in an optimized way. As ARPANET began through the involvement of the NSF to evolve into the Internet for general use, the steady stream of ideas became a flood of techniques to submit, control, and schedule jobs across enterprise systems. The latest in these ideas are the grid and cloud, intended to be used by a wide variety of different users in a non-hierarchical manner to provide access to powerful aggregates of distributed resources. Grids and clouds, in the ideal, are intended to be accessed for enterprise computation, data storage and distribution, and visualization and display, among other applications without undue regard for the specific nature of the hardware and underlying operating systems on the resources on which these jobs are carried out. In general, grid and cloud computing is client-server computing that abstract the details of the server away—one requests a service (resource), not a specific server (machine). However, both terms are vague from the point of view of computing architectures and computing models and referring to "everything that we already do".
The reality at present, however, is that service-centric grids are still very difficult for most users to access, and that detailed and low-level programming must be carried out by the user through command line and script execution to carefully tailor jobs on each end to the resources on which they will run, or for the data structure that they will access. This produces frustration on the part of the user, delays in the adoption of enterprise techniques, and a multiplicity of specialized “enterprise-aware” tools that are not, in fact, aware of each other which defeats the basic purpose of the grid or cloud. Most enterprise programs are still not written in metaprogramming languages but line by line in conventional compiled languages or scripting languages as decades ago. The current state of the art is that these conventional programs and scripts can provide instructions of a metaprogramming language with a relevant operating system handling the service grid as its metaprocessor. However, there are presently no acceptable metaprogramming methodologies to program, deploy, and dynamically federate these services for a metaprogram securely and efficiently with fault detection and recovery. In this presentation the emerging protocol neutral SORCER using service-object oriented architecture with its exertion-oriented networking and its exertion-programming model is contrasted with the service protocol-oriented architecture (e.g. OGSA, CORBA, RMI) that limits us to one fixed wire protocol.
Mike Sobolewski received his Ph.D. from the Institute of Computer Science, Polish Academy of Sciences. He is the Principal Investigator of the SORCER Lab (SORCERsoft.org) focused on research in distributed service-centric metacomputing. Currently he is a World Class Collaborator at the Air Force Research Lab (AFRL), WPAFB/USA and a Visiting Professor at the Polish Japanese Institute of IT, Warsaw, Poland. Before, he was a Professor of Computer Science, Texas Tech University and Director of SORCER Lab from 2002 till 2009. Now he is engaged in development of Algorithms for Federated High Fidelity Engineering Design Optimization applying his innovative SORCER solutions at AFRL.
While at the GE Global Research Center (GRC), 1994-2002, he was a senior computer scientist and the chief architect of large-scale projects funded by the United States Federal Government including the Federated Intelligent Product EnviRonment (FIPER) project and Computer Aided Manufacturing Network (CAMnet). Also, based on his web-based generic application framework he developed seventeen successful distributed systems for various GE business components. Before his work at GE GRC he was a Research Associate at the Concurrent Engineering Center (CERC) and and Visiting Professor at Computer Science Department, West Virginia University (1998-1994). At CERC/WVU he was a project leader for knowledge-based integration for the DARPA Initiative in Concurrent Engineering (DICE).
Prior to coming to the USA, during 18-year career with the Polish Academy of Sciences, Warsaw, Poland, he was the Head of the Pattern Recognition and Image Processing Department, the Head of the Expert Systems Laboratory, and was engaged in research in the area of knowledge representation, knowledge-based systems, pattern recognition, image processing, neural networks, object-oriented programming, and graphical interfaces. He has served as a visiting professor, lecturer, or consultant in Sweden, Finland, Italy, Switzerland, Germany, Hungary, Slovakia, Poland, Russia, and USA.