Showing papers on "Database-centric architecture published in 1993"

Merging application-centric and data-centric approaches to support transaction-oriented multi-system workflows

Abstract: Workflow management is primarily concerned with dependencies between the tasks of a workflow, to ensure correct control flow and data flow. Transaction management, on the other hand, is concerned with preserving data dependencies by preventing execution of conflicting operations from multiple, concurrently executing tasks or transactions. In this paper we argue that many applications will be served better if the properties of transaction and workflow models are supported by an integrated architecture. We also present preliminary ideas towards such an architecture.

Task Decompostiion Through Competition in a Modular Connectionist Architecture: The What and Where Vision Tasks

Abstract: A novel modular connectionist architecture is presented in which the networks composing the architecture compete to learn the training potterns. An outcome of the competition is that different networks learn different training patterns and, thus, learn to compute different functions. The architecture performs task decomposition in the sense that it learns to partition a task into two or more functionally independent tasks and allocates distinct networks to learn each task. In addition, the architecture tends to allocate to each task the network whose topology is most appropriate to that task. The architecture’s performance on “what” and “where” vision tasks is presented and compared with the performcmce of two multilayer networks. Finally, it is noted that function decomposition is an underconstrained problem, and, thus, different modular architectures may decompose a function in different ways. A desirable decomposition con be achieved if the architecture is suitably restricted in the types of functions that it can compute. Finding appropriate restrictions is possible through the application of domain knowledge. A strength of the modular architecture is that its structure is well suited for incorporating domain knowledge.

Analyzing the Properties of User Interface Software

Abstract: Software architecture is an increasingly important research topic and in this report we investigate the potential role of architecture in evaluating the properties of a system built to a particular architecture. Currently such architectural analysis is complicated for two main reasons: authors of new architectures describe their creations in idiosyncratic terms; and there is no clear way of understanding an architecture with respect to an organization''s life cycle concerns -- efficiency, maintainability, modifiability, and so forth. This report addresses these shortcomings by proposing a domain-based method for analyzing software architectures called SAAM (Software Architecture Analysis Method). This method contains several steps. A canonical functional partitioning for the domain is adopted. Next, some candidate architectures in this domain are described in a common and simple structural language, providing a neutral context in which to understand their similarities and differences. Next, life cycle concerns for the quality of the resultant software are determined and a set of benchmark tasks are created which embody these concerns. Finally, the architectures are evaluated and compared with respect to how well they support the benchmark tasks. This report illustrates the method by analyzing user interface architectures with respect to the quality of modifiability.

The influence of the object-oriented language model on a supporting architecture

Abstract: The MUSHROOM architecture is designed to efficiently support dynamic object-oriented languages such as Smalltalk. The authors describe the architecture, and how the object-oriented model of computation and the provision of a sophisticated computer led to various simplifications in the design and implementation of the architecture. Two particular features of the object-oriented model have been exploited in the architecture. Freedom of object identifier selection, allows for straightforward and efficient implementation in software of virtual-to-real address translation and for efficient object allocation, and the absence of address aliasing allows simple, virtually addressed caches to be used. A prototype implementation of this architecture is currently being commissioned. >

Control system architecture: the standard and nonstandard models

Abstract: Control system architecture development has followed the advances in computer technology through mainframes to minicomputers to micros and workstations. This technology advance and increasingly challenging accelerator data acquisition and automation requirements have driven control system architecture development. In summarizing the progress of control system architecture at the last International Conference an Accelerator and Large Experimental Physics Control Systems (ICALEPCS) B. Kuiper (1991) asserted that the system architecture issue was resolved and presented a "standard model". The "standard model" consists of a local area network (Ethernet or FDDI) providing communication between front end microcomputers, connected to the accelerator, and workstations, providing the operator interface and computational support. Although this model represents many present designs, there are exceptions including reflected memory and hierarchical architectures driven by requirements for widely dispersed, large channel count or tightly coupled systems. This paper describes the performance characteristics and features of the "standard model" to determine if the requirements of "non-standard" architectures can be met. Several possible extensions to the "standard model" are suggested including software as well as the hardware architectural features. >

Structuring distributed shared memory with the pi architecture

Abstract: The authors examine the use of a new architecture for the design of system software. The architecture allows the construction of flexible system software components, and the resulting realizations can be tailored to the needs of various applications. The focus is on the application of that architecture to distributed shared memory (DSM). The system software architecture, called pi , defines elements of system software as generalized objects. These x objects have two interfaces: one for usage and one for additional control. pi enables system software components to evolve as computer hardware evolves and it addresses the demands emerging classes of applications like object-oriented databases and multimedia. The pi architecture and its use in a DSM subsystem for clustered workstations are discussed. >

A Massively Parallel, Micro-grained VLSI Architecture

Abstract: This paper describes the design and implementation,al aspects of a high performance micro-grained architecture with enhanced communication capability. This architecture is capable of teraops performance. The architecture is organized as a semi-systolic array of processors. A prototyping system for the architecture which will provide control, I/O, and an interface to a host system for the micro-grained architecture is specified. Beyond the research outlined here, the prototyping system can be used as a ‘?est-bed” for various clnss/siirdent VLSI design projects. This architecture will be useful for solaing a number of important problems, such as: edge detection, locating connected components, two-dimensional signal and image processing, sorting elements, and performing element permut a t i o 71.s.

Advanced function printing: from print to presentation

Abstract: The strength of Advanced Function PrintingTM (AFPTM) is due largely to the architectures that form its foundation. The architectures on which AFP is based have been developed over the last 12 years and have influenced the development of standards, competitive architectures, and, most importantly, software inside and outside IBM. Customers are demanding a more comprehensive view of printing that includes easy creation, viewing, and even specialized editing of printable documents. These "next generation" requirements are now being satisfied by software products that are based on the existing architecture. This paper describes some of these products and how they use the architecture, and describes possible future directions for AFP and related technologies.

Darts: a domain architecture for reuse in training systems

Abstract: The dynamics involved in the training system marketplace of today are dictating the need for major changes in the way organizations specify, develop, and maintain training systems. One of the key areas affected by these changes is the system and software architecture of training systems. This is evidenced by the increased attention that has been placed on architectures by recent initiatives (e.g. Structural Model, Mod Sim, STARS, DIS, ARPA DSSA, etc.). There are many reasons for this emphasis, not the least of which is a desire to produce training systems at the least possible cost while providing faster time to market and higher quality. An architecture for training systems can be a framework to enable cost reduction, reusability, and standardization. We derive a set of attributes which we believe characterize a "good" software architecture. We discuss an architecture developed by Boeing Defense & Space Group, the Domain Architecture for Reuse in Training Systems (DARTS) and evaluate DARTS against these criteria. We also discuss the role of DARTS in megaprogramming, part of the ARPA STARS initiative, and suggest that DARTS is a suitable architecture for achieving the STARS vision of process-driven reuse.

A hybrid electro-optical architecture for classification trees and associative memory mechanisms

Abstract: An architecture is presented here which can be used for some important paradigms of intelligent systems. This architecture targets applications which require real time processing of stream inputs with versatile hardware which exploits parallelism. The architecture is particularly suited for pattern recognition paradigms which are based on the use of decision trees. Artificially intelligent systems based on decision trees interestingly present some common computational requirements which can be served very efficiently by a Data Flow architecture. A small set of different functions is computed repeatedly with simple result tokens passed from one computation to successive ones. Developments in optical processing have introduced elements which are particularly suited to the computational requirements of some of these systems, and therefore they can be effectively employed in this architecture. The architecture presented here is based on Data Flow design principles and is enhanced with optical processing elements. The function of the architecture is illustrated by discussing the mapping of two specific AI paradigms—a pattern classifier and an associative recall mechanism.

Adapting Plan Architectures

Abstract: Publisher Summary This chapter discusses the planning with the behavior of large real-world systems. In real-world planning, the actions in the plan serve as instructions that cause a system to behave in a certain way. The objective of planning is to get the system to behave in a way that solves some problem and solves it well. The ARCHON approach is to incrementally modify the initial architecture to improve the system's ability to produce good plans. The utility of this approach depends on the extent to which the modification process can be controlled and the extent to which it is independent of the details of any particular architecture. ARCHON allows architecture modification only in terms of template-based fixes. This is a self-imposed restriction, allowing the modification process to be controlled to whatever extent is necessary. The internal consistency requirements of the architecture limit the occasions on which a proposed modification is allowed—the ramifications of a proposed modification must also fit into the architecture. The architecture modification approach is, therefore, controlled by deliberate restrictions and by the natural stability of the environment.