As identified above the fully developed architecture and methodology would have to contain the following deliverables:
In its report and subsequent article [1] the IFIP/IFAC Task Force identified two possibilities to develop the required complete architecture and methodology:
Both technically and strategically the combination of existing architectures and methodologies has great advantages and this line is followed from here on. However, the strategy to develop GERA is not necessarily the same as the strategy to develop GEEM or the other deliverables.
The respective strategies should be acceptable to the community of users and developers which in turn depends on the following criteria:
It is possible to give separate consideration to each of the deliverables respecting in each case the above strategic requirements.
The development of the Generic Enterprise Reference Architecture (GERA)
The building of GERA should be similar to building a car-race track and defining the rules of the game rather then building a Formula-1 T-model of enterprise engineering. GERA is an open system framework.
GERA must be a public domain document (possibly considered for standardisation) to allow meaningful and comparable competition to occur in the area of enterprise engineering. Similar goals were adopted by projects as CIM-OSA and the existing links to standards bodies (CEN-CENELEC, ISO TC184 etc) must be built upon as well as links to other groups involved in enterprise modelling (ICEMT SIG).
Examples of candidate competitive products which could arise out of GERA, or could use GERA as a leverage are: Computer-aided Enterprise Engineering, Enterprise Modelling and Project Management tools; Integration Platforms; Enterprise (re)Engineering Know-how; Educational Material etc.
Milestones to develop GERA:
(i.e. including the concept, identification and
requirements and design of GERA). The specification will
also have to include the definitions of the
other deliverables.
Presentation of GERA for the user community:
For the user community, preferably use the two tiered diagram format (now presenting four such diagrams according to the four matrices). The two tiered diagram format will group together those ``subject matter'' areas which will belong together in the methodology. This presentation can be prepared on the basis of the PERA document by adding further chapters.
The second presentation is for the methodology-, modelling-, and standardisation community, preferably in the form of the four matrices. The matrix presentation will be seen as a further development of the CIM-OSA cube. The matrix presentation will use a refined subject matter presentation drawing on the one hand on the corresponding CIM-OSA definitions and on the above subject matter groups (as developed for the user community).
The Development of a Generic Enterprise Modelling Technology (GEMT) and
Generic Enterprise Models (GEM)
The development of GEMT is not independent of the most generic enterprise models (GEM) because these are the meta-models for the language design (the ontological theories define the semantics of the language). For this reason the two are dealt with together.
The GEMT is a collection of modelling tools (languages / templates) which cover the needs of GERA. The definition can be done on two levels.
The first level is the separate definition of suitable modelling languages for each area. This is expected to be done for each area in GERA, at least for those subject matters which require model development in the application phase of the generic methodology (minimum requirement).
The second level is where the meta-models of these modelling languages are expressed as views of a common meta-model. If this is done then far more possibilities exist for enterprise engineering tools to analyse, execute (e.g. simulate), cross check, and validate models than in the first case. This integration of languages is also necessary for a smooth transition from the design to the building and operating of enterprise models.
The second case also allows for the adoption of the General Modelling System and Permissive Modelling concept [4], which are based on the recognition that designers need to be able to use a wide variety of languages to express themselves. Once a core set of languages (CIM-OSA) is expressed as views of the underlying meta-model, it will be possible for other views to be developed as add-ons. Information contributed to the enterprise design database using either the core languages or the add-on languages will all be added to the underlying design database and thus becomes visible through the defined views.
A major part of necessary languages was developed in the CIM-OSA project and in GIM (organisational/decisional modelling), and some additions will have to be considered (e.g. temporal behaviour, cost/economies). In other words, the modelling technology of GERA should be an open system as well.
The granularity of modelling is a problem at the moment. CIM-OSA strives at model executability down to the level of enterprise operation. This requirement limits CIM-OSA's ability to incorporate in the modelling language anything more then a simple ontological theory of events/actions and data (for efficiency reasons).
TOVE also strives at executability but is less concerned with the efficiency of execution. For this reason a complex ontological theory of actions, time, cost etc. is part of the meta-model.
The tension between the two approaches is apparent and a possible solution is
proposed here. However, further investigation is necessary in this regard.
It is proposed to express CIM-OSA modelling languages as views of a more complex ontology (TOVE) than the CIM-OSA
meta-model. In this way a design system based on CIM-OSA models could take
advantage of the added semantics of CIM-OSA concepts, and the
improved analysis capability in the design phase. However,
for efficiency, the simpler, compatible, CIM-OSA meta-model
could form the basis for building and operating enterprise control software.
For the user of GEMT there would be no difference seen, but for the developer of enterprise engineering tools and integration platforms the differentiation would be visible.
Milestones to develop GEMT and GEM:
Given the above uncertainties the milestones need to be re-evaluated in light of the feasibility of the first few steps.
Up to this point this is a feasible engineering project. The following tasks, however, need to be addressed by research.
One suitable expression of the modelling concepts of the languages so defined is the definition of the basic language concepts as modelling classes in a single object-oriented language[6].
This step need not necessarily lead to the design of any new language (although some extensions may have to be done both on the language and the meta-model level.)
The Development of a Generic Enterprise Integration and Engineering Methodology (GEEM)
It is acknowledged that more then one methodology may exist
to cover the GERA framework. The present
proposal points out how one such methodology can be developed
with relatively small expenditure, drawing upon the Purdue and GIM methodologies.
Other, equally viable proposals could be presented on the basis
of only one of them or based on other combinations with proprietary
methodologies.
The first goal of developing GEEM is to provide one complete methodology which covers the entire GERA architecture. The coverage of the methodology may be achieved by a./ adding more information to the Purdue Implementation Procedures Manual b./ adopting compatible methodologies which complement the Purdue Implementation Procedures Manual with methods in the areas not addressed (or not addressed in sufficient detail).
The second goal developing GEEM is to specialise the methodology such that a consistent set of modelling languages is proposed along with the presentation of procedures. This will allow the addition of examples more easily. The second goal supposes that the GEMT project provides the necessary input. However, because the basis of that is the CIM-OSA modelling languages, this is not a heavy constraint.
Milestones to develop GEEM: