Task 1120: Distribution of Product Data Models - of the VEGA project has developed the deliverable D102 in two parts.
The major goal of this task is to define potential elements and guidelines of a methodology for the distribution of STEP model instances. This methodology tries to take into account the different possible levels of granularity for the STEP models distribution, and rely on specific IT concepts for distribution as it will be implemented into the Corba Access to STep models (COAST : VEGA - WP3).
Development around distribution of STEP models is not intended as a modification of the current Product Modelling technology or standards, but as a transparent extension to provide distribution control information and a third path (besides SPF files and the SDAI) for access to STEP-based information. Even if the methodology developed here is related to STEP, this is not additional concept to STEP and EXPRESS, or a new EXPRESS flavour: this is why we will talk, as far as possible in that deliverable, about distribution modelling (rather than EXPRESS-D, as initially characterised in the VEGA Technical Annex). The same is also valid for EXPRESS-W which will be referenced as Workflow meta-model (see D103b Bakkeren ).
The first part of this deliverable, D102a, defined the scope of a model for distribution of STEP data, in relation to existing standards for modelling and distribution. It described what could be the main features of this model and gave some first hints on the different levels of granularity of the STEP distributed data. Moreover, D102a promoted the elaboration of an EXPRESS-D language (as initially identified in the VEGA TA): the main goal of such a language would have been the specification of the availability of data through their distribution, using the EXPRESS language as a basis, with annotation (or pragma instruction) extension, allowing simple addition of information to existing EXPRESS schemata for distribution purpose, and offering an upward compatibility with EXPRESS-IS. This language would have authorised, in particular, the distribution of any entity object at the level of the middleware.
The identification of what seems to be the best level of object distribution granularity for the COAST (distribution of STEP models) and consequently the giving up of EXPRESS-D (model and language) are the main results of the current second part of this deliverable (D102b). D102b first identifies in more details the distribution of STEP models and data, constituted by what is directly visible on the network and managed by the COAST. The distribution of product data (defined by EXPRESS entities) may be the cause of dramatic overheads over the networks depending on the type of application (but especially in Building & Construction CAD design, where thousands of similar data may be defined when designing a building under a CAD tool), and this is the major reason why the object layer chosen in VEGA for distribution and management by the COAST is the model: the designer of a COAST based distributed application will have to specify which models of the application have to be distributed (and thus data which will be remotely accessible) and which models are internal to a specific server and not accessible.
Distribution of STEP models nevertheless is sometimes not enough when dealing with real industrial applications: D102b also investigates specific notions as "grouping of entities", views or inter-model relationships to meet the specific requirements of the LSE virtual enterprise, in line with needs identified in D501a (Liebich ), mainly for performance goals as a mandatory objective for network-based architecture like the COAST. Thus, this deliverable specifies the distribution and associated properties of STEP data objects as it could be further integrated into the COAST architecture. These concepts will be then implemented and evaluated in the VEGA project, especially in the course of the VEGA WP3 Task 3200 - Implementation of the COAST.
The distribution of STEP models and data is one of the main objectives of the VEGA project. Data distributed over a network are potentially available for any customer, thus data have to be protected through access rights to control who access the data, and moreover must be accessed in the right time and at the right place: distribution features are the ground for Security and Workflow considerations. The distribution in VEGA relies on the concept of STEP model, as intended in the ISO 10303 standard, a set of related entity instances based on an EXPRESS schema definition, and has to be related to inter-model relationships and schema interoperability. Moreover, the granularity of distribution greatly impacts the way followed by clients to access data through the middleware, and therefore influences the definition of a Query Language for the access and modification of product data distributed across the network. As previously told, the COAST is the VEGA distribution layer, and this deliverable is also an input for the specification and implementation of the COAST, which will include a set of optimised services for the distribution of STEP data. Finally, product data as well as administrative data like documents for instance, will be accessible through the COAST, following the main guidelines identified hereinafter. Consequently, this deliverable, which is the result of Task 1120, can be considered as relevant to the following tasks: Task 1220/1230 - Workflow and Security Modelling supporting tools Task 2100 - Schema Interoperability Task 2200 - Functional Interoperability Task 3100 - Specification of the COAST Task 3200 - Implementation of the COAST Task 4100 - Specification of Distributed Information Service - and also to the different implementation tasks of Work Package 5 leading up to the final demonstrations of the VEGA project.
The VEGA TA stated that the main goal of task 1120 should be an EXPRESS-D model and language for a neutral description of the distribution of STEP data based on any object broker, and D102a tried to apprehend what could be the main objectives and elements of such a model, and mainly identifying the following points handling and availability of the potential (different types of ) distributed objects on one hand, the levels of access rights and access protection for the distributed objects on the ther hand. EXPRESS-D meta-model and language are no more considered in this second part of D102, for reasons introduced above and mainly exposed in section 3 of the current deliverable. Consequently, EXPRESS-D will be no more referenced in that deliverable except in section 3.3 for specific purpose revealed in the section itself.