Enterprise Architecture – from Blueprints to Design Services F. Lillehagen, D. Karlsen, H. G. Solheim, H. D. Jørgensen & H. Smith-Meyer Troux Technologies AS, Lysaker, Norway B. Elvesæter & Rolf Kenneth Rolfsen SINTEF Information and Communication Technology, Oslo, Norway ABSTRACT: This paper looks at the evolution of Enterprise Architecture modeling and use, since its birth in the late 1970’s, then looks at the role of architectures in leading companies today, and finally developments to add more value to past investments are presented and discussed. Until recently enterprise architecture projects have been focused on business and systems information gathering and presentation, primarily providing document services to managers. With new advances in active knowledge modeling and in integrated service architectures many initiatives are started addressing all enterprise knowledge layers. The main purpose of these initiatives is to integrate active knowledge models, improve view handling, provide model-driven usercomposed services, and support dynamic architecture driven solutions. Finally new approaches to innovative design, whether the focus is on product, organization, process, system or new enterprises are presented. Enterprise Architecture is not just about technology, so in the future we must include social and human aspects and integrate the architecture of the human mind and develop new forms of organizational structures. 1 INTRODUCTION Enterprise Architecture (EA) can be defined as: a simplified and aggregated representation of many layers of enterprise knowledge, defining forms, features, roles and functions. EA solutions should embrace architectures and views of many enterprise knowledge spaces and dimensions, and must support many perspective views. For IT people EA is a plan which shows the main features and characteristics of the system that will be analysed, justified and agreed before detailed technical design can commence. Architectures should be shared and discussed enterprise-wide between all users and stakeholders [7, 14]. Enterprise architecture is both a challenging and confusing concept. For decades, construction industry has used architecture in the design and construction of all size buildings. Their “architecture” utilizes standard symbols, norms and rules that can be recognized and understood by all members of their industry for carrying out the construction work. The systems engineering community by comparison has never had the advantage of this type of “time-tested” structure. Instead, since the beginning, many heterogeneous architecture proposals have been developed. They are often overlapping approaches and the underlying concepts are not explicitly defined. Simi- larities and differences between architectures can not be perceived nor annotated by users. This creates obstacles for its correct understanding in industry and finally its acceptance and use. Architectures of products, organizations, processes and systems, and operational solutions can not at this point in time be adequately described, analyzed, used and managed by designers and engineers as a coherent solution-space to meet their challenges. 1.1 Motivation and background Enterprise architectures are today developed to give managers a better understanding of the things the enterprise owns, operates, and produces so they can make better business decisions. Likewise, enterprise planners need architectures so they can understand what they have, what they don’t have, and what they need to have to fill the gaps. For IT managers the features and functions to be provided by hardware and software systems are often described in architecture models in such a way that the proposed systems can be evaluated for how much they contribute to increased effectiveness of operations. This information and knowledge is used to make better decisions about how best to spend the IT budgets. Enterprise architecture, considered as the foundation of enter- prise systems, has emerged as a ‘tool’ to help stakeholders manage system engineering and IT change. But EA is not only an IT issue, but influences business, organizations, products, and strategies. In spite of all the great things that architectures can provide to business managers, IT people, enterprise planners and stakeholders, there are still severe shortcomings in the architecture modeling approaches and techniques. Many architecture frameworks have been developed to address different aspects of architecture development, but frameworks can not support design and team-working [18]. 1.2 EA blueprint approaches Most current EA approaches are information gathering, filling in views that are used as blueprints to document the legacy [10]. Examples of such frameworks are: Department of Defense Architecture Framework (DoDAF) [1], Treasury Enterprise Architecture Framework [2], The Open Group Architectural Framework (TOGAF) [3], Generalized Enterprise Architecture and Methodology (GERAM) [4] and the Zachman Framework [5]. Shortcomings of these frameworks are that they tend to be documentation-heavy or process-heavy, they do not provide operational support; they do not provide customer knowledge services nor support knowledge activation, and they are not suitable to smallmedium-enterprises (SMEs) [18]. 2 THE EVOLUTION enable new approaches to model-designed solutions and model-configured systems [8, 18]. The support of mental, social and new organizational models complementing the technical models and matching models in the human mind has long been a challenge. The first attempts at modeling of human behavior and our minds were made as early as 1980 to help design good graphic user interfaces. But we had to wait ten years for more scientific approaches, ref. Senge’s work on mental models [13]. Once mental and social models can be expressed and represented, prototypes could be available around 2008, we will be able to develop environments that support integrated design and learning. 2.1 The layered Enterprise Architecture An important step was taken when the layered Enterprise Architecture was introduced (Lillehagen 2002). The layered architecture, introducing the active knowledge layer, was the basis for the IDEAS Framework, which has later been adopted by ATHENA and other projects [7]. Exec tasks BOA EKA Business models BPM MUPS EKA Services The evolution of Enterprise Architecture (EA) started with the publication of the Zachman Framework in 1978 explaining the complexity of enterprise information management. But the market break-through came with the technology of Enterprise Modeling. A major market push was provided by the US Congress in 1996 when it passed the Klinger-Cohen Act demanding that all public organizations develop and document their enterprise architecture as a prerequisite to invest in IT. POPS SW Repos. serivices ICT MDA SOA ASA COTS Figure 2: The layered Enterprise Architecture with the core services, perspective views and many knowledge aspects at each layer. Most abbreviations used should be well-known. 1978 1996 2005 2012 2020 Zachman Klinger- Knowledge Social Design & Framework Cohen Act Models Models Learning Figure 1: The evolution of Enterprise Architecture from 1978 until today, and the evolution for the next 15 years. This year we are witnessing the introduction of true active knowledge modeling and representation at the heart of architecture models delivered as integrated modeling and execution platforms. This will The layers are denoted the business operational layer, the active knowledge layer and the ICT layer. The active knowledge layer separates the increasingly dynamic business behavior from the software, and supports user interactive involvement to change the models and solutions at any time. The main contents of the layers are identified in figure 2. The Business Operations layer (BOA) has a kernel of business process management and execution capabilities, including business documents, rules and protocol handling. This is complemented by a set of standardized tasks and task patterns for performing types of business and business models. The Enterprise Knowledge layer has the POPS language at its core supported by the EKA services, then other EKA services are used to extend and evolve the POPS language meta-models to cover also the needs of other knowledge spaces. 2.2 Integrating Reference Models There are some 20 industrial application area reference models being worked on with the purpose to standardize data exchange and information flow, Examples are models for supply-change management implementing the SCOR methodology, and for configuration management and CMMI methodology. The groups that develop reference models have little or no contact, so the models will not be compliant nor coherent. This is potentially a major source of non-interoperability. These reference models must be integrated in the EKA by appropriate metamodels and modeling languages as extensions to the POPS language. This is illustrated by the six white cylinders illustrating the EKA services in figure 2. The EKA layer affords responsible users services to use Model-configured User-composed Platform Services (MUPS) to build and manage their own services and solutions. The MUPS approach to solutions development and management promises to provide trust and confidence in use of services. 2.3 Operational Architectures The AKM approach to modeling has since its inception been looking for ways of developing integrated modeling and execution platforms supporting the development of architectures as part of operational solutions. The ATHENA project has focused on developing the Modeling Platform for Collaborative Enterprises (MPCE). The MPCE is the first version of such an integrated platform [16]. 2.4 Design Support With the introduction of integrated modeling and execution platforms, the POPS language and the EKA services, we will have a platform capable of supporting design. To enable true design the MPCE must support iterations, view handling, concurrent alternatives and eventually parameter trees, and provide a setting for social models supporting teambuilding and knowledge-sharing. 3 THE ARCHITECTURE OF THE MIND Recent research in modern sciences, such as phenomenology, evolutionary psychology and cognitive neuroscience (Gazzaniga, 2002) indicate that knowl- edge is managed mainly, if not totally, as views in our mind [15]. This means that good user workplaces and working environments should support powerful view handling and that we have to understand that complementary views, views containing different perspectives of the same scene or artifact, may be partly in our minds and partly in the visual enterprise models. This also means that it is hard to judge by looking at external views: what exactly is knowledge. A view of data or information to one person can be knowledge to another person. It all depends on what complementary views you already have access to in your mind. What we can say is that in order for anybody to have knowledge at least three complimentary views must be associated. 3.1 Knowledge Existence Four zones of knowledge existence, perception and representation are described and illustrated below. (a) Horizon of Externalization – expert knowledge, internal and tacit (b) Horizon of Participation – participative knowledge, social and external (c) Horizon of Comfort – sensed knowledge, our real world precept (d) Horizon of Reason – general acquired knowledge d c b Horizon of Extern a Horizon of Particip. Horizon of comfort Horizon of reason Figure 3: Depicting the four horizons of knowledge existence emphasizing the importance of proximity, medium and encoding language in representing knowledge. Each zone is “delimited by a perceptive horizon or border, defining knowledge existence, form and value as dependent on proximity to action and method of knowledge encoding and diffusion.” The horizons define the edges of four knowledge spaces of human minds as illustrated below. “Enterprises have many knowledge spaces shared among many humans; the innovative, the operational, the governing, the evolutionary, the strategic, the cultural, and a multitude of smaller cascaded spaces.” The four main enterprise knowledge spaces that we are beginning to understand how to represent are the innovative, the operational, the governing and the evolutionary. Each of these spaces has four main dimensions represented by different aspects depending on the purpose of the model. Different aspects of the same dimensions and domains may be replicated in many knowledge spaces. a Informal knowledge Derived knowledge Participative knowledge Situated knowledge Action knowledge Figure 4: Knowledge within the areas bounded by the four horizons of knowledge existence will vary in detail, form, actuality and timeliness. The closer to the scene of action the more value and richness has the knowledge The most valuable knowledge is the work generating and generative knowledge. In figure 4 this is labelled Action knowledge (centre cylinder), The second most valuable is the knowledge acquired by participative actions. These two spaces are referred to as “Situated knowledge”, and is the core knowledge we capture and harness when people build teams and develop shared active models. Situated knowledge is mostly pragmatic logic, and semantics is primarily added to provide identification and navigational properties (nomenclature). Knowledge in the two outer spaces can also be of great value, but in a business network focusing on core business: core or situated knowledge and related data are vital for success. Situated knowledge has four intrinsic properties of great importance for human perception, design and learning. These are called reflection, recursion, repetition and replication, and are explained in many publications and papers [9]. (b) Accessibility—the market penetration and familiarity of the representation language throughout industry and among software professionals. (c) Usability—the ease with which new users can adopt and become proficient in the use of the modeling and representation language (d) Expressiveness—the ability of the representation language to capture unambiguous meaning, logic and semantics relevant to a given domain (e) Life cycle coverage—the scope of the representation language’s utility throughout the development life cycle (e.g., design, architecture, implementation, requirements, testing) The evaluation criteria can be aggregated into a figure of merit for each domain language being considered and evaluated. If this was done for all modeling tools and methodologies in the market, then very few would qualify as appropriate EKM languages. EKM languages and tools also have to be supported by platforms closing the gaps between knowledge modeling and management, and execution of work. 3.3 Models and Views An active knowledge model can be much richer in knowledge content than a model of real-world artifacts or products. Most knowledge models comprise real-world objects and layers of active knowledge, represented as structures and views. If knowledge models are created and owned by a person, a designer-team or an innovative enterprise, then they express the experiences, pragmatic logic and externalized tacit knowledge of leading practitioners. Other knowledge models may be created and built by participative persons or novelties that do not express the same multitude of views as they do not own the core knowledge. The latter models will naturally not be as rich in content, and the views may not be as detailed. Model ownership, practitioner tacit knowledge externalization and pragmatic logic and experience capture is key to active knowledge modeling. 3.4 View Orientation 3.2 The Knowledge Modeling Approach In order to select the best Enterprise Knowledge Modeling (EKM) approach for active model development, services enabling MUPS, and model execution, we need to define evaluation criteria. The key criteria are: (a) Processability—the ability of the representation language to be efficiently processed by programs tasked with inferring truth values and providing interoperability. Enterprise knowledge modeling is to a large extent about developing or generating views of different spaces, aspects and perspectives, and representing them in a knowledge base for direct reuse by role performers, or as a base for other derived views or even as base for new models. There are many types and kinds of views, and view ownership by roles. Types of views can be defined based on various principles and criteria, and be characterized and form categories The need to structure views in a logical way is apparent in order to be able to give meaningful support to users. We have found that it is useful and correct to differentiate between knowl- edge defining, representing and presenting types of views and kinds of views (role views) with the meanings described below. Knowledge defining views are of these four main types and categories: - Descriptive views - Context views - Functional views - Content views A descriptive view is a view that contains metadefinitions, a view that is used as language definition for other views. The reason for calling it a descriptive view rather than a meta-view is that industrial practitioners do not think in terms of meta- and instance views. It is a lot easier for them to understand the term descriptive, in the meaning that one view can act as a description of another view. A context view is a view that is used to define the context required by a functional view, to be able to make the functional view fully specified. A functional view is a view that is used to define the details of a given function/activity as e.g. a process model that describes all the execution tasks and GUIs of a process. Diagrams are other examples of functional views A content view is a view that contains operational information and data – the domain data. A type of view specifies the design, the purpose and the behavior of a view, and must be supported by a technical implementation. Kinds of views are used to configure the types of views according to a purpose or a role that is going to use it. It could be a management view or a development view of the same enterprise object that use the same type of view but looks different and shows different content. The different content may be controlled by data access and/or by preferences, while the view style may be controlled by style sheets or other means. The authors have worked to find scientific evidence and principles as a base to implement support for improved view management, and will later be reporting on their findings. 4 ENTERPRISE KNOWLEDGE SPACES An enterprise has many Enterprise Knowledge Spaces (EKS). These spaces will be implemented as Enterprise Visual Scenes (EVS) by developing and applying the Active Knowledge Modeling (AKM) technology and realizing that we have to be trained to think and work in multiple dimensions [8]. Visual scenes will transform the World Wide Web into the innovative, high-value multi-medium envisioned. EVS, and model-driven working environments will augment the capabilities of our senses and minds. This will have great impact on the future of computing, and on most sciences by enabling human minds to extend the use of the visual powers of the brain. Currently less than 10% of the capacity of the lefthalf of the brain is activated by current learning and training methods, and by our ways of working [15]. The four main types of knowledge spaces we have to better understand and support for future enterprise design and development of visual scenes are briefly defined as being: The Innovative Space The Operational Space The Governance Space The Evolutionary Space These four types are each bounded by four major knowledge dimensions. The description of the dimensions and their domains and types of views is work ongoing among the authors, and will also be reported on in future papers. 4.1 The Innovative Space The Innovative space supports creative work (invention, reuse, design, implementation, validation and learning). Its main concepts are the Enterprise Knowledge Spaces (Industrial War-room principles) implemented as Enterprise Visual Scenes, and implemented by the POPS (Process, Organization, Product, System) core modeling language and methodologies. The innovative space enables and manages continuous change and evolution in product, process, organizational and system structures of the total Enterprise Architecture modeled and captured in the Enterprise Knowledge Architecture (EKA). 4.2 The Operations Space The Operations space supports business and work operations (plan, assign, operate, generate, adapt, extend, manage and terminate). Its main concepts are the Continuous Business Solutions (CBS) generation and the Visual Enterprise Computing (VEC) delivery approach, supported by the MUPS approach and multiple life-cycle management (adapting and extending the intelligent infrastructure), implemented as steps in the VEC delivery methodology: PoC – EA - E I/ EAI - Solutions Generation, User Deployment and Agility Management. 4.3 The Governance Space The Governance space supports work progress monitoring and regulatory compliance (govern, monitor, measure, decide, change, experiment and strategize). Its main concept is bubble-models, aggregation and propagation of parameters, attributes and values, the “real-time enterprise”, implemented as services to explore and investigate, to support decision-making, manage and govern value-creation in parallel life-cycles, need for powerful analysis and interaction techniques beyond balanced score-cards, use of "value-metric charts" and other value matching techniques. 4.4 The Evolutionary Space The Evolutionary space supports life-cycle performance management and improvement (analyze, configure, compare, align, transform, and manifest). Its main concepts are continuous Collaborative Business Management (CBM) and Business Transformation, implemented as supporting many concurrent Collaborative Business Solutions (CBS’s), concurrency is dependent on collaborative models and EKA services. A new form of networked organization is being developed, defining service-teams owning certain core services, like modeling, training and other categories of basic services. These teams will dynamically and continuously develop and deliver solution services supported by adaptive and extensible EKA structures created by executing EKA services. If the strategy and need is to rapidly generate new executable solutions, then first time modeling may not costs so much, but if the strategy is to develop a model supporting innovative work or transforming business practices then the first time investment may be huge. The pay-off should be comparative. The first project to develop a knowledge model may be a major investment, as it is important to adopt a stepwise process and secure the widest possible involvement. The costs will also depend on customer maturity in most technical, social and nontechnical areas. The second or third time a similar modeling project is performed with the same or a similar customer the amount of modeling will be much less. Figure 5 illustrates that the amount of modeling is dependent on the No. of users and their intended use of the model, on the processes in focus and on the purpose and ambition of modeling. Models across many companies and different industries should be developed as and composed of sub-models, so sub-models and views should be easily reused and adapted from model repositories. 5 USING ACTIVE KNOWLEDGE MODELS The knowledge modeling approach is part of an overall architecture development process, and it has to involve the practitioners. Successful models, models that have value to users actually depend on involving practitioners in the modeling, and working in teams. This requires a proven approach, coherent methodologies, integrated modeling and execution platforms, and capabilities to integrate the legacy. 5.1 The Extent of Modeling Many people involved in enterprise modeling and enterprise knowledge modeling (in our mind two different disciplines) are skeptical to the extent of work involved and the costs of modeling. With active knowledge modeling, based on the MPCE and applying the MUPS approach, industry will find modeling effective, beneficial and boundarybreaking. No.of users 1st Proces- 2nd 3rd Purpose of model Figure 5: The amount of modeling first time can be costly and time-consuming, but less so than systems development, and the benefits can be manifold. 5.2 The Importance of Customized Approaches and Methodologies Many modeling projects fail to deliver to expectations because the project is not properly anchored in the organizations of the participants and the influencing stakeholders. There is not just one approach to model an enterprise. The best approach evolves with the knowledge of the other dimensions. An example of a very successful architecture and solution modeling project is the NOSA project on Earth Science Research [13]. The approach taken in this project is simple but has worked wonders. It is therefore worth a study. 6 INTEGRATED ENTERPRISE SERVICE ARCHITECTURES In the ATHENA Integrated Project [16] partners are developing an Integrated Enterprise Service Architecture (IESA) addressing the shortcomings of today’s enterprise architecture frameworks. The IESA describes the technical architectures of an integrated modeling and execution platform that allows us to describe different views of an enterprise using active models. These models can be explicitly associated with software services, and not just static documents and diagrams. This approach enables the enterprise architecture to be an operational tool that can be used to ensure consistency and manage dependencies and trade-offs between various concerns in the business decision-making process. 6.1 Social and non-technical services The ATHENA and INTEROP NoE [19] projects have adopted a holistic perspective on interoperability based on the IDEAS framework. However, this framework does not cover the non-technical aspects of an enterprise. The framework defines layers of technical architectures and views where each layer focuses on particular activities, structures, and services. The framework suggests that interoperability must be achieved on all layers within and between cooperating enterprises. • Interoperability at the business layer can be seen as the organizational and operational ability to cooperate and share knowledge and values. • Interoperability at the knowledge layer can be seen as the compatibility and the coherence of common and private knowledge assets. • Interoperability at the ICT layer can be seen as the easy with which software components can be exchanged, changed and developed at runtime. (Advanced Technologies for interoperability of Heterogeneous Enterprise Networks and their Applications Integrated Project) [16]. The work does not represent the view of the European Commission or the ATHENA consortium, and the authors are solely responsible for the paper’s content. 9 REFERENCES [1] [2] [3] [4] [5] 6.2 Model-generated workplaces [6] Model-generated workplace must be part of any working environment for knowledge workers. The ability of any architecture to model, generate, use, monitor, manage and reuse workplaces in different life-cycles is decisive for enterprise efficiency and effectiveness and thus for return on investment. 7 CONCLUSIONS AND FUTURE WORK [7] [8] [9] [10] Enterprise Architecture is developing from being a complex mosaic of static information views to become an integrated Services Architecture supporting mental and social models and model-driven design. Modeling approaches, languages and methodologies, platforms and solutions are being worked on to enable concurrent design and development. The four main types of Enterprise Knowledge Spaces and their implementation as visual scenes are concepts that need to be further developed and validated. The description and expression of spatial dimensions and their domains and types and kinds of views is work ongoing among the authors. The results are promising, and will be reported on in the Athena project reports and papers in 2006. The major contributions of our research are the development of an active knowledge modeling technique for use in the development of operational and evolving enterprise architectures and solutions. 8 ACKNOWLEDGEMENTS The work published in this paper is partly funded by the European Commission through the ATHENA IP [11] [12] [13] [14] [15] [16] [17] [18] Department of Defense, "DoD Architecture Framework", Department of Defense Architecture Framework Working Group, 2003. Department of the Treasury, "Treasury Enterprise Architecture Framework, Version 1", Department of the Treasury, CIO Council, 2000. The Open Group, "The Open Group arcitectural framework (TOGAF), Version 8", The Open Group, 2002. http://www.opengroup.org/architecture/togaf8/ IFIP-IFAC, "GERAM: Generalized Enterprise Architecture and Methodology, Version 1.6.3", IFIP-IFAC Task Force on Architectures for Enterprise Integration, 1999. ZIFA, "The Zachman Institute for Framework Advancement", 2005. http://www.zifa.com Lillehagen, F. & Krogstie, J. 2002, Active Knowledge Models and Enterprise Knowledge Management, in Enterprise Inter- and intra-organizational integration, Kluwer Academic Publishers, Kosanke K. et al (ed.) Lillehagen, Frank: From Enterprise Modeling to Enterprise Visual Scenes, paper presented at the CE 2003 conference, Madeira, July, 2003. Lillehagen, Frank: “The Foundations of Active Knowledge Modeling technology”, paper presented at CE 2003, Madeira, July, 2003. Lillehagen, F, Enterprise Modelling and AKM technology, ICEIMT, Verdal 1997 Lillehagen, F. & Krogstie J. 2001. Active Knowledge Models and Enterprise Knowledge Management, ICEIMT workshop, Paris, December 2001 Karlsen. D, Lillehagen F. Solheim H.G and more: “ The AKM Intelligent Infrastructure. Paper to be presented at the CE 2004 conference in Beijing, July 27th 2004. Senge, P. M.: The Fifth Discipline; The Art and Practice of the Learning Organization., Doubleday Currency, New York 1990. Martin, James, N.: Using an Enterprise Architecture to Assess the Societal Benefits of Earth Science Research, paper to be presented at INCOSE, Rochester, July, 2005. Martin, James, N.: Dr. Dissertation Proposal; Adding Knowledge Modelling Capabilities to DODAF., Dissertation tbs to George Mason Univ., US, March 2005. Gazzaniga, Michael: "Minds Past" Book buplished by Barnes and Noble, see www.barnesandnoble.com, ISBN 0520224868, published July 2004 ATHENA, "ATHENA Home Page", ATHENA IP, 2005. http://www.athena-ip.org/ ATHENA A1, "D.A1.5.1: MCPE Specification, Version 1.0", ATHENA IP, Deliverable D.A1.5.1, August 2004. https://www2.mycomputas.com/Team/Repository/Project s/Project_204/Upload/Attachments/WPA1.5/DA_1_5_1_ MPCE_Specification_final.zip ATHENA A1, "D.A1.1.2: "Stae-of-the-art in EM and EAl", ATHENA IP, Deliverable. will be made avaialble to the public soon. [19] Interop NoE, "Interop NoE Home Page", Interop NoE, 2005. http://www.interop-noe.org/