Next: 3 The RASSP Authorization Model Up: Appnotes Index Previous: 1 Introduction

Reuse Methodology and Implementation Appnote

2.0 Architecture of the RASSP Enterprise System

Figure 2 - 1 shows the data flow architecture of the RASSP enterprise system. A design engineer interacts with a workflow manager to perform various predefined process steps. On execution of a particular process step, the workflow manager may invoke the appropriate CAD tool for the activity, opening the requisite product data files and making them available in the designer's workspace. The workflow manager interacts transparently with the product data manager on behalf of the user to open, close, save and configuration manage the product data files. The user may also interact directly with the desktop manager to invoke tools outside of the workflow environment, as needed

Figure 2 - 1: Enterprise System Data Flow Architecture

Once a CAD tool has been invoked, the designer interacts with the tool in its native environment to perform a design or analysis activity. He or she may invoke the RRDM through the CAD tool environment or directly. The RRDM may receive reusable design knowledge from the CAD tool, or meta-data only in cases where the design is managed in place. Additional meta-data is solicited from the environment as well as from the designer through the use of domain-specific templates. Both the meta-data and design knowledge (where applicable) are stored in the local design knowledge repository. The workflow manager for the RASSP Enterprise System is implemented through Intergraph Corporation's Design Methodology Manager (DMM), and the product data manager is implemented by the Intergraph Asset information Manager (AIM). Additional information on the RASSP Enterprise System is provided in Enterprise Application Note

The architecture of the RASSP Reuse Data Manager (RRDM) and its relationship to users and other system resources is detailed in Figure 2- 2. The RRDM consists of a client browser, a "perspectives" application server (including the search and vocabulary mapping engines), and multiple source data repositories. A designer looking for reusable design knowledge browses the hierarchy and identifies a potential source for designs of interest. Then, through a custom view of the common vocabulary, the user may specify attributes of the candidate design(s) he or she is interested in. Queries may include arithmetic expressions, logical expressions, attributes of related classes, and constraints on any of the attributes, as appropriate. Mapping algorithms that account for a variety of data inconsistencies, differences in precision, translation of units of measure, and other distinctions among terms across repositories are resolved to the user's customized vocabulary and display requirements. The search of available (and optionally selected) source repositories is performed in parallel, with meta-data for all results displayed through the client browser. Depending on the number and nature of the results returned, he or she may further refine the search by specifying additional attributes and/or more restrictive criteria. Objects of interest may be selected and the corresponding designs (e.g., schematics, drawings, models) displayed in their native tool environments given that appropriate resources are available to the user.

A prototype of the RASSP Reuse Data Manager (RRDM) was demonstrated in April 1998, providing access to multiple, distributed repositories of VHDL models. The key capabilities featured at this RASSP demonstration are summarized below.

• Data Modeling - Through the use of the Ontolingua tool, the actual ontology components developed in support of the RASSP satellite communications and electronics domain were highlighted. These components share basic information-asset characteristics, such as the date of creation, version, author, and so forth, but are classified uniquely by fit, form and function on the design object hierarchy, and by the nature of the information (e.g., VHDL model, script, test vector, specification) in an engineering data object hierarchy. For the most part, multiple engineering data objects were collected to define a single design object. The resultant ontology components have been published and made available through the Stanford Knowledge Systems Laboratory (KSL) Ontolingua server.

• Search and Retrieval - The ontology components were implemented and demonstrated in the prototype environment, showing hierarchy and relationship browsing as well as attribute-based query construction and results retrieval. Physical design objects related to the meta-data were also displayed in their native environments. Examples of designs used for demonstration purposes include complex architectural, behavioral, and performance models developed for the RASSP benchmark program, with the relevant documentation, source code, test environments, and drawings where available. Also included were a number of VHDL models developed for the program by several universities and South Carolina Research Authority (SCRA).
• Heterogeneous Integration - Integration of heterogeneous sources (i.e., with syntactically and semantically distinct database schemas) was demonstrated through the use of DBMS vendor tools in addition to the RRDM. A complex query across multiple sources was initiated from the RRDM, and the results examined. Each individual repository was then searched using vendor-provided tools to show not only that there were distinctions in the data stored in each repository but in the terms defined by the repository schemas as well.
• Distributed, Object-Oriented Architecture - The demonstration environment included multiple client browsers hosted on several machines (both PCs and UNIX workstations), multiple back-end repositories hosted on distinct server machines, and a single perspectives server host. The commercial product will support multiple, distributed application servers in order to optimize local performance in addition to distributed clients and source repositories. The prototype provided source repository selection capabilities, which, in contrast with physical examination of individual repositories, demonstrated distributed repository access capabilities.

Planned Enhancements

A number of additional capabilities are planned for the commercial product and subsequent releases by Sandpiper Software, Inc. These include:

• Function-Based Search and Retrieval
• Content-Based Search and Retrieval
• Configuration Management for all ontology components and meta-data with optional release management for design knowledge
• Fine-Grained Authorization Control
• Enhanced User Authentication
• Extensive User Customization Capabilities, including user and role-specific views of the virtual repository as well as support for custom forms, templates, and reports
• Standards-based application integration for third-party tools through CORBA-compliant APIs, including automated data synchronization where required
• Multi-vendor DBMS support (relational, object-relational, and object databases)
• Intelligent agent integration for data mining and decision support, subscription, notification, and other extensions to the baseline technology

Operational support tools and utilities (e.g., ontology generation and development, data migration, system administration) are also being developed independently by Sandpiper Software to support the commercial information broker product.

Next: 3 The RASSP Authorization Model Up: Appnotes Index Previous: 1 Introduction