ArchiMate. ArchiMate Made Practical. Modeling according to ArchiMate guided by a collection of good practices

Transcription

1 ArchiMate ArchiMate Made Practical Modeling according to ArchiMate guided by a collection of good practices

2 ArchiMate Colofon Title : ArchiMate Made Practical Date : 01 April 2013 Version : 4.0 Change : Translation of Dutch 4.0 version which added new good practices. Status : Final Editor : Hans van Drunen, Egon Willemsz Company : Workgroup ArchiMate Usage/Tools Authors : Harmen van den Berg, Alexander Bielowski, Gertjan Dijk, Hans van Drunen, Gert Faber, Jan van Gijsen, Rienk de Kok, Ger Oosting, Wim Schut, Frans Smit, Jan van de Wetering, Egon Willemsz Prior verslons: Hans Bosma, Frank Langeveld, Joost Luijpers, Robert Slagter Translation: Harmen van den Berg, Aad-Jan Binneveld, Erik Borgers, Hans van Drunen, Rienk de Kok; with a special thanks to Sandy Britain for reviewing the English translation ArchiMate is a registered trademark of The Open Group

3 ArchiMate Table of Contents Colofon 2 Table of Contents 3 1 Introduction Motive Goal and Intended Public Reading Guide Changes since version Website Call for new good practices 3 2 Modeling in the business layer Overview of good practices Business service, business function and business process Demarcating business services Demarcating business processes Demarcating business functions Business process and business interaction Viewpoints for process modeling Collaboration between organizations Information domain 13 3 Modeling in the application layer Overview of good practices Application landscape Interfaces between applications Functionality of an application Websites as an application component and the usage of services 21 4 Modeling in the infrastructure- / technology layer Overview of good practices Infrastructure services Infrastructure landscape Difference between system software and artifact 25 5 Modeling over the boundaries of layers Overview of good practices Most commonly used relationships 27

4 ArchiMate 5.3 Product and application services Support of batch and online processes Service broker Domain Internal and external services Necessity for the usage of services Simplification and preserving consistency Grouping Information exchange with external organizations Display messages in queues Nesting and implicit relations File transfer 51 References 52

5 1 Introduction 1.1 Motive How do I model an application landscape? How do I relate logical (implementation independent) and physical (implementation dependent) business processes? How do I match the application portfolio with the technical infrastructure? How do I visualize what data are required when delivering which products and services to customers? How do I model a service broker? These are all examples of modeling questions that an enterprise architect will be confronted with. There are many variants of modeling solutions that are being practiced in order to answer these questions. These variants lead to inconsistencies and can limit good communication. Behind every model there is an original story, but a model can often evolve into its own new truth where the original story will get lost over time, especially if the originator of the model is no longer involved in the maintenance of that model. Companies that practice working under architecture are using ArchiMate more and more as a descriptive language for capturing enterprise architectures consisting of domains such as: products/services, processes, organization, data, applications and technical infrastructure. Using this language it is possible to model the structure and behaviour? within a domain as well as the relationships between domains. Another important aspect of ArchiMate is the possibility of visualizing models using multiple viewpoints, specific to different stakeholders. Besides this, the language provides a formal foundation to enable analysis of models (for example on an aspect such as performance). ArchiMate is deliberately not intended to replace existing languages like BPMN, UML etc.; it is positioned to deliver value in addition to these existing languages. The intention is to tie in as much as possible with existing standards and practices. The ArchiMate language is currently undergoing widespread adoption by an increasing number of Enterprise Architects both in the Netherlands and Internationally. Whilst this has been beneficial in providing a level of standardisation to the models created by architects, what we can observe is that the same architecture problem can be modelled in multiple different ways, and some ways modeling practices, techniques and conventions are more effective than others. Thus those architects who are skilled and experienced in the use of the language are able to produce models with greater accuracy, consistency, clarity and communicative value than the majority of newcomers. This guide aims to capture a number of effective practices, distilled from experience, in modelling common architecture problems using ArchiMate. 1.2 Goal and Intended Public The ArchiMate workgroup has received a considerable number of questions about how ArchiMate concepts should be applied to perform real and practical work. A R C H I M A T E M A D E P R A C T I C A L 1

6 The aim of this document is to start answering these questions by describing proven solutions, documented as good practices. This will lower the threshold for applying ArchiMate in practice and will increase the effective value of the developed models, and consistency in approaching common modeling situations This document is written for enterprise architects who want to apply ArchiMate in practice as a language for capturing models of organizations, the ICT-support and the relationships between them. This document should be treated in addition to existing ArchiMate documentation (see the appendix for references): Enterprise Architecture At Work. Archimate 2.0 An Introduction Archimate 2.0 Reference Cards. Viewpoints Functionality and Examples. ArchiMate 2.0 Specification 1.3 Reading Guide The following chapters contain a collection of good practices. For each documented good practice, the following structure is used: Name in the paragraph title: a recognizable name that summarizes the contents of the good practice Question: a description of the question that is leading to the development of the good practice Solution: a description and visualization of the preferred solution that gives an answer to the question Consequences: a description of the consequences of the solution, for example in relation to communication with stakeholders, consequences for implementation, etc. Alternatives: a description of alternative solutions with the justification for each alternative Relationships with other good practices: a description of the relationships that exist with other good practices In this document, two types of good practices have been collected: Good practices that give an answer to conceptual ArchiMate questions (how does one use a certain concept in practice?). Good practices that give an answer to modeling questions. The good practices are organized as follows: Modeling within the business layer: chapter 2. Modeling within the application layer: chapter 3. Modeling within the infrastructure/technology layer: chapter 4. Modeling across boundaries: chapter 5. A R C H I M A T E M A D E P R A C T I C A L 2

7 1.4 Changes since version 3.0 This document is a renewed version of the booklet ArchiMate in de praktijk version 4.0 published in 2012 by the working group ArchiMate Usage/Tools. In this version 4.0 new best practices have been added. In addition text has been modified for readability and the style of figures are made more consistent. The new good practices are: Information domain Nesting and implicit relations File transfer 1.5 Website More information about the ArchiMate Forum, the activities of the working group ArchiMate Usage/Tools and the collected good practices can be found on the websites: and On the last site, new good practices will be published before these will be integrated in a next version of this booklet. 1.6 Call for new good practices The content of this document is by no means complete and leaves questions unanswered. This document is therefore designed to be a living document. It contains a collection of good practices harvested from day-to-day usage. Our intent is to get the reader started with these good practices, and that, based on this usage, new insights will emerge which will be subsequently integrated into this document. Do you have any modeling questions that you would like answered? Do you have good practices that you want to share with your peers? Or do you want to participate in the working group ArchiMate Usage/Tools, where we collect and discuss good practices and integrate them into this document? You can contact the chairs of the working group via the address: h.vandenberg@bizzdesign.nl en egon.willemsz@uwv.nl Or by telephone, contact Harmen van den Berg ( ) or Egon Willemsz ( ). A R C H I M A T E M A D E P R A C T I C A L 3

8 2 Modeling in the business layer 2.1 Overview of good practices In this chapter, the following good practices are described: 1. Business services, business function and business process. 2. Demarcating business services. 3. Demarcating business functions. 4. Demarcating business processes. 5. Business process and business interaction. 6. Viewpoints for process modeling. 7. Collaboration between organizations 8. Information domain 2.2 Business service, business function and business process Question: In the business layer, the concepts business service, business function and business process are positioned. The exact difference between these concepts is not clear in all cases, or to be more exact, in practice there sometimes exists confusion if something needs to be tagged as a business service, a business function or a business process. In the next section we give definitions and descriptions of these concepts. Solution: A business service represents the added value that an organization delivers to its environment. One can make a distinction between internal and external services: Internal services represent the added value that is being delivered within the domain that the service belongs to. External services represent the added value that is being delivered to other domains or to the environment (for example customers). A business function is an area that the organization wants to pay attention to (e.g. by putting energy into, structurally committing resources to etc) in order to support its business goals. A business function can therefore be positioned as a grouping of internal behavior based on certain criteria (for example location (same department), communication, required skills, shared resources and shared knowledge). A business function represents a part of the added value of on organization. A business process is a unit of internal behavior or a collection of causal (sequence or dependency) related units of internal behavior, with the goal of producing a pre-defined collection of products and services. A business process can be constructed from sub processes. A business process is triggered (started) by one or more business events or other business processes. Informally one could say that a business process consists of a number of activities or subprocesses that are being executed in a certain sequence. Every activity is part of a business function. In other words, a process combines a chain of activities each of which are part of business functions, as depicted in the figure below: A R C H I M A T E M A D E P R A C T I C A L 4

9 Figure 2-1: Example of business functions in relation to processes A single process will not always belong to a single business function (as in this example): a business function will almost always consist of multiple activities and process steps and a process will often be realized by multiple business functions. Business functions as well as business processes describe the internal operations of an organization. A business service describes the parts of business processes and functions that are externally visible and usable. It describes the service that can be used, without necessarily describing the details of how that service is realized by means of processes or business functions. In the figure below, the most commonly used relationships between these concepts are visualized (being a part of the ArchiMate meta model): Figure 2-2: Most commonly used relationships between concepts Consequences: Not applicable. Alternatives: Not applicable. Relationships with other good practices: The distinction that exists between business service and business process / business function is also valid between application service and application function. Comparable criteria and heuristics can therefore also be applied to the application layer. A R C H I M A T E M A D E P R A C T I C A L 5

10 2.3 Demarcating business services Question: In the business layer, one of the concepts is the business service. But how does one distinguish a business service and how is it being demarcated? Solution: If something is tagged as a business service, than that service should in principle be consumable by multiple consumers. Services can be grouped by means of aggregation relationships. This implies that it is only sensible to disaggregate a service into partial services if these partial services can also be consumed independently of each other. Apply the following heuristics to distinguish business services: Distinguish a service from the perspective of its consumer. The service should be recognizable and usable for the consumer. The naming convention should be done from the perspective of the service consumer. Distinguish services based on the activities that are executed in the business layer, and based on the products that are being delivered. Model the services from the outside in: defined by their use. Distinguish different services optimized to support different concerns. Prevent overlap in offered services: different services offer different behavior. Overlap in behavior is an indicator that you need to define the overlapping behavior as a separate service. A service is realized by one or more business functions or processes that represent concrete internal behavior of the organization. A business function can realize multiple business services. Always model which business functions or processes a business service realizes and which business processes (in case of internal services) consume the service. An internal business service is always used by at least one business function or business process Keep services consistent: make sure that comparable behavior is offered as services in a comparable manner Use services to hide implementation details. It is sufficient for a service consumer to know that a certain service is being offered and how the consumer must use the service. A service consumer does not need to know how a service is realized. Consequences: Not applicable. Alternatives: Not applicable. Relationships with other good practices: Identical criteria and heuristics can also be used for application services in the application layer and for infrastructure services in the infrastructure layer. A R C H I M A T E M A D E P R A C T I C A L 6

11 2.4 Demarcating business processes Question: In the business layer, one of the concepts is business process. But how does one distinguish a business process, and how is it being demarcated? Solution: Apply the following heuristics to distinguish business processes: A business process is triggered by a business event, and ultimately delivers a service or product to a customer, or partial products or partial services that are used as part of a service or product for a customer. Divide a process into phases that can be being treated sequentially. Examples of phases are request, handling and after-sales. A phase often coincides with a process or function within the organization. A frequent pattern of phases is the value chain. One can recognize phases by the assignment of actions to different actors, combined with a timing sequence between the actions. Group actions based on the time that they happen e.g. online- or batch processing, daytime or night-time). Divide a process into parts based on the knowledge and skills that are required to execute certain actions. This can be deduced from the functions that are tagged for each task. Divide a process into parts based on function segregation or other forms of internal measures. Divide a process into parts based on geographical boundary (physical location) where the activities take place, for example a region. Divide a process into sub processes that can be executed independently. This can occur where multiple (partial) products are being delivered. Distinguish partial processes that occur more than once. These common partial processes can be generalized and re-used. Distinguish partial processes that are repeated as a whole. Consequences: Not applicable. Alternatives: Not applicable. Relationships with other good practices: Not applicable. A R C H I M A T E M A D E P R A C T I C A L 7

12 2.5 Demarcating business functions Question: In the business layer, one of the concepts is business function. But how does one distinguish a business function and how is it being demarcated? Solution: Apply the following heuristics to distinguish business functions: Make a separate function for the interactions with each environment actor. For example; create a function for suppliers, consumers and government. This heuristic has the following specializations: Separating relationship management from other functions: Make a distinct function for activities that deal with customer contacts. The customer here is the external customer of the company, but for large companies it might also be applicable for an internal customer (for example another department). Separation to market: Make a distinct function for each customer segment/target group of the company. Examples are business customers and private customers. Make distinct functions for processes with different kind of triggers (business events). In this way processes can be identified that have periodical triggers (monthly receipt of declarations) and that have incidental triggers (quotation request). A specialization is: Case distinction: Make distinct functions for activities that are related to different cases. A company can make distinction between different damage claim cases: damages less than or greater than 1000,-. In this example, a function for a small and a function for a large damage claim case could be distinguished. Make a distinct function for each product group, for example for activities related to damage insurance versus other insurance. A distinction can also be made centered around any business object: Distinction related to business object: Make distinct functions for a group of activities, if they all work on a certain business object. This way, functions can be distinguished that are related to invoices, cash flows, offers, customer history etc. Make a distinct function for each phase or state that a product can be in. In case of a damage insurance claim this could be, for example, distinct functions for requesting, reward raising and handling of damage claims, closing and management. Make a distinct function for a group of activities, whenever the activities require special 1) skills, 2) expertise or 3) responsibilities. Examples are judicial knowledge, actuarial expertise, communication skills or authorizations for certain decisions. Make distinct functions for activities that control the primary process, (planning / control). Make distinct functions for activities that change the primary process or its implementation. This leads for example to distinct functions for marketing, product development and system development. Model functions inside-out: defined by their 'capability'. Consequences: Not applicable. Alternatives: Not applicable. Relationships with other good practices: Comparable criteria and heuristics can also be used for application functions on the application layer. A R C H I M A T E M A D E P R A C T I C A L 8

13 2.6 Business process and business interaction Question: Besides the business process concept, ArchiMate also understands the concept of the business interaction. A business interaction is a process that is executed by two or more actors (a collaboration of actors). Why is the business interaction concept needed on top of the business process concept? Isn t it sufficient to relate a process to a collaboration of actors and thereby express that the process in itself is in fact an interaction? Solution: A business interaction is defined as a specialization of a business process (or more precisely a specialization of the generic behavior concept). By using a business interaction, information is added to more explicitly describe the behavior that is executed by a collaboration of roles or actors. By relating a business process to a business collaboration the same information is expressed and it more explicitly describes the behavior that is executed by a collaboration of roles or actors. In the following figure, some variants are expressed: Figure 2-3: Variants of business process and business interactions In the top left, a business process is modeled that is assigned to two roles, in the top middle a business process is assigned to a business collaboration, bottom left an interaction is assigned to a business collaboration, bottom middle shows a business collaboration consisting of two roles and to the right an interaction that is assigned to a business collaboration consisting of two roles. This redundancy is not without a good cause. In some cases, a business collaboration is not explicitly modeled (for example if the focus is on the behavior and not on the actors), and at the same time you want to express that it is about an interaction with multiple roles. Another example is if one would want to express a collaboration, but not make explicit which parties or actors the make up the collaboration; this can be useful in, for example, a development phase where it is not yet clear who the final parties will be that ultimately will add value to a product of service. A R C H I M A T E M A D E P R A C T I C A L 9

14 Consequences: The goal the architect has with modeling or visualizing a business interaction or collaboration will to a large extent determine which modeling solution is chosen and which aspect gets focus. Alternatives: Not applicable. Relationships with other good practices: A comparable distinction as described here is applicable to the application layer, especially related to application interaction and application collaboration. The examples and heuristics shown here are also applicable. A R C H I M A T E M A D E P R A C T I C A L 1 0

15 2.7 Viewpoints for process modeling Question: How do you model a process in ArchiMate? The ArchiMate book shows a number of viewpoints but which combination of viewpoints is required to model a process containing sequential steps? Solution: The ArchiMate book shows two viewpoints for process modeling: the Business process collaboration viewpoint and the Business process viewpoint. The difference between these viewpoints is that in the Business collaboration viewpoint the relation between processes and the relation with the environment is visualized, while in the Business process viewpoint one business process is expressed. We concentrate here on the Business process viewpoint; the Business process collaboration viewpoint is analogous. When modeling a process in ArchiMate, the following aspects can be visualized: the parts of the process, their cohesion, the assignment of process parts to roles, actors of business parts and the support by applications. This can be realized in the following sequence: 1. Model the business process at a high level: Decompose the business process and make relationships by introducing business events and triggering relationships. One can add to that the way in which the process delivers business services to its environment. 2. Information-exchange: Model the information-exchange between processes by adding flow relationships between processes, or reading and writing of business objects. 3. Assignment to organization: Model the organizational part that executes the business process. This can be visualized by assignment relationships or by using swim lanes. 4. Support by applications: Model how applications support business processes. The application viewpoint can be used for this. If required, a more generic distinction can be made between a logical and concrete or physical level of modeling. A logical model expresses the functional structure and logical cohesion of business processes. No detail about time, place, people or machines is given. An implementation model will closely resemble what you can see or want to realize in reality. It describes physical aspects like humans and machines, with concrete details assigned to tasks. If an organization wants to distinguish between different types of processes (like logical and physical) in ArchiMate, such distinction should be defined before starting modeling. The ArchiMate relationship used between logical process and physical process is composition. A process can be specialized into process types. For example, the process Request insurance can be specialized into the sub processes Request travel insurance and Request damage insurance. Consequences: Not applicable. Alternatives: Not applicable. Relationships with other good practices: Refer to the good practices Business service, business function and business process and Constraining business process. A R C H I M A T E M A D E P R A C T I C A L 1 1

16 2.8 Collaboration between organizations Question: How can we model collaboration between organizations? What is meant by business collaboration? How does business collaboration relate to the concepts of business interaction and business role? Solution: For a proper insight into collaboration between organizations, the Co-operation Viewpoint is of value. This viewpoint addresses how the various roles and actors work together in a value chain. Business Collaboration: A temporary- or permanently collective of two or more business roles that work together and therefore interact. The business interaction concept is the behaviour we see as a result of business collaboration. Business actor: An active entity performing behaviour. An actor can equally be a single person,e.g.. customer, or a group of people e.g. a business line. Business role: The responsibility for performing specific behaviour, to which an actor can be assigned. The view below provides insight about the roles in the value chain and how they collaborate. Figure 2-4: Co-operation between organizations Consequences: Not applicable. Alternatives: Not applicable. Relationships with other good practices: See also good practices: - demarcating business services, business processes, business functions - Business process and business interaction. A R C H I M A T E M A D E P R A C T I C A L 1 2

17 2.9 Information domain Question: In our organization it is important to have a clear framework of concepts. To do this I would like to use an information model (business object model). How can I do this with ArchiMate modeling and what types of relationships can I use between the objects? Solution: We can illustrate the answers to these questions with the fictitious example below. : The business rules we wish to describe are as follows: An Employment object holds information about the relationship between an employee and an employer. Based on the Employment the Employee is obliged to be insured in the context of a certain Social Security Law. In addition, a non-worker can insure themselves on a voluntary basis. Within the Employment there are Working Periods which are distinguished by constant properties (e.g. wage data). A simple model based on the description above is as follows: Figure 2-5: High-level information model The relationships between business objects in this variant are indicated by associations. Associations are the weakest and most general form of relationship in ArchiMate and serve only to show that there is a relationship between the business objects. The example above is therefore semantically weak, though not incorrect. The reader can determine that; there are associations between the business objects, but what they mean is not expressed. A model such as that shown above is eminently suitable for communication with stakeholders who have little or no knowledge of data modeling, such as stakeholders from the business or from management. They are also useful when modeling at an intermediate stage in which the information model is still being developed but the precise relationships between the information object is not yet known. Consequences: Not applicable. Alternatives: a more detailed variant looks as follows: A R C H I M A T E M A D E P R A C T I C A L 1 3

18 Figure 2-6: Detailed information model The relationships between business objects are styled by stronger forms of relationship types. This gives the model more meaning. In the example, the following relation types are used: Aggregation: The employment is an aggregation of an employee and an employer. Composition: The employment contains working periods. Specialization: An employee is a specialization of a natural person as well as the insured. Realization: The employment contract is a realization of an employment In ArchiMate, it is not possible by default to add cardinality and optionality to these relationships. Specific domain languages (such as UML) are meant for that. Such properties can, however, be added in some tools. In situations where the target audience is familiar with data modeling, and/or when there is a need for a more precise definition of the relationships between information objects, this second variant is better suited than the first variant. It is of course also possible to derive the first variant from the second variant in case we need to do so for communication purposes, since it is only the level of abstraction (generality) that is being altered in the view. The pattern of associations remains identical. Thereby the simplified view can be displayed, even while it is under-pinned by the more detailed model. Relationships with other good practices: Not applicable. A R C H I M A T E M A D E P R A C T I C A L 1 4

19 3 Modeling in the application layer 3.1 Overview of good practices In this chapter, the following good practices are described: 1. Application landscape. 2. Interfaces between applications. 3. Functionality of an application. 3.2 Application landscape Question: My application landscape is very complex, how can I get grip on it, how can I model it? An overview of applications is often visualized in an application landscape. This is done to provide an overview of the relationships between applications. This will often produce a large number of connecting lines and actually no overview at all. This kind of overview often only shows how complex the application landscape is. It does not give any assistance in application portfolio maintenance. Solution: It is important to keep overviews readable. Apply a rule of maximum 10 to 15 applications per overview. If there are more, it is recommended to use a general model containing only a major grouping and a detailed model per group. Choose a viewpoint that is important for the stakeholder you are addressing. The format will always be an Application structure viewpoint. A classification can be done via process, organization, data, type of functionality or infrastructure. Group applications per area of interest and only connect the areas. This helps to reduce the number of lines in a diagram. With an increasing number of applications, it will give more of an overview if relationships are visualized in a matrix. Below an example is shown of an application landscape organized by application type (process management, input processing, processing, output processing and complete process support). A R C H I M A T E M A D E P R A C T I C A L 1 5

20 Figure 3-7: Application landscape organized along type of functionality of applications Below an example is shown of an application landscape according to usage pattern by front-office and back-office. Figure 3-8: Application landscape organized along usage of front-office and back-office Consequences: For an overview of all relationships you need multiple views. Use a matrix for this purpose. A tool can be very valuable in this case to register information and produce a relationship matrix. Alternatives: The alternative is the large view with all applications and the many lines. This is only usable to express how complex the landscape is and does not give much additional added value. Relationships with other good practices: There is an important relationship with Interfaces between applications A R C H I M A T E M A D E P R A C T I C A L 1 6

21 3.3 Interfaces between applications Question: It is desirable to visualize interfaces between applications. Which ArchiMate concepts can be used for this purpose? Solution: An application is modeled as an application component with corresponding application functions. An interface between applications is modeled via an application service. The application service delivers data by means of an access relationship with a data object. In the example model application function A realizes an application service that is used by B. The application service delivers the data that are in the data object. Figure 3-9: Application interface using application functions, application service and connected data object Consequences: If existing interfaces between applications in a company are visualized according to this pattern, it might appear to suggest that a service oriented architecture is in place while in reality this does not have to be the case. Alternatives: With compliance to rules of composition (refer to good practice Simplification with sustainable consistence ), application functions can be left out of the model. The figure below shows the result. Figure 3-10: Application interface using application service and connected data object The above interface can also be modeled without a data object: A R C H I M A T E M A D E P R A C T I C A L 1 7

22 Figure 3-11: Application interface using application service Not every organization will have a service oriented architecture. In that case it is also possible to model application interfaces using a relationship between the applications. This can be done using for instance a used-by relationship (denoting that Application A is used by Application B) or by a using a flow relation (denoting that information flows from Application A to Application B, or from B to A). In the example below: Application A delivers customer data to application B. Figure 3-12: Application interface using a flow relationship In some cases interfaces between applications are not realized directly, but via a database. One application writes into a table that is read by another application. Below figure shows this. Application A writes, Application B reads the same data object. Figure 3-13: Application interface using data object Relationships with other good practices: This can be used by constructing an application landscape using the good practice Application landscape. It is also applicable in conjunction with with the good practice Service broker. A R C H I M A T E M A D E P R A C T I C A L 1 8

23 3.4 Functionality of an application Question: How can you model externally visible functionality of applications with ArchiMate? Solution: The ArchiMate concept that was designed explicitly for modeling externally visible functionality is the application service. The application service is connected with a realization relationship to an application function belonging to an application. Corresponding interface(s) can be visualized if required. The figure below shows this modeling pattern: Figure 3-14: Externally visible functionality This way of modeling can be used both for internal and external application services. Consequences: If an application produces a lot of functionality, a model like this can become cluttered by the great number of realization relationships. The suggestion can be easily induced that a service oriented architecture is in place while in reality this is not so. Alternatives: This model pattern can be simplified by leaving out the application function and deduce the realization relationship to the application component (figure below to the left). If the interface has no relevance for the model, it can be left out (figure below to the right) Figure 3-15: Externally visibly functionality - simplified When no services can be distinguished, the externally visible functionality of an application can be visualized by delivering the application function via an interface to the consumer. In the example below, a Business process is supported by an Application function, which is exposed by an Application interface. A R C H I M A T E M A D E P R A C T I C A L 1 9

24 Figure 3-16: Externally visible functionality coupled to a process role In the last view the application interface can also be left out. The result is a used by relationship between the application and the business role. Relationships with other good practices: For functionality that is not interactive there is a relationship with good practice Interfaces between applications. Visualizing the functionality of an application resembles visualizing applications within an application portfolio. The good practice Application landscape is also applicable for the functionality of an application. A R C H I M A T E M A D E P R A C T I C A L 2 0

25 3.5 Websites as an application component and the usage of services Question: The functionality of back office and legacy applications is increasingly disclosed via web applications. The functionality of the back office application is often used in multiple websites, and a website can also use external services. How to model these situations? Solution: The situation is explained using the following example: Figure 3-17: Websites and application services The back office application realizes a service 'premium calculation'. This service is used by the two websites. Back office functionality is thus disclosed for these websites. Website 2 is using two external services: 'Authentication' (e.g. DigiD) and 'Route calculation' (e.g. using GoogleMaps). Consequences: Not applicable Alternatives: Not applicable Relationships with other good practices: This good practice is related to the good practice Interfaces between applications A R C H I M A T E M A D E P R A C T I C A L 2 1

26 4 Modeling in the infrastructure- / technology layer 4.1 Overview of good practices In this chapter the following good practices are described: 1. Infrastructure services. 2. Infrastructure landscape. 3. Difference between system software and artifact 4.2 Infrastructure services Question: Which services would one want to model on the infrastructure layer? From a maintenance point of view the infrastructure layer often requires a lot of detailed information. For example which systems exist and how are these connected to each other? The step in describing meaningful services for the environment is not commonplace: how does one start with modeling services on the infrastructure layer and which types of services must be distinguished? Solution: Describe those services on the technology layer that provide support to applications. It is recommended to distinguish processing services, storage services and communication services. When modeling the infrastructure layer it is essential to use abstraction of internal details, for example implementation details. Domain specific languages are better suited for this. Use the viewpoint Infrastructure usage from the ArchiMate book (Lankhorst et al., 2005, p. 187). This viewpoint allows visualizing how applications are supported by software and hardware infrastructure. Figure 4-18: Example of infrastructure services Apply the definition as described in ArchiMate (Lankhorst et al., 2005): an infrastructure service is a visible unit of functionality, delivered by one or more nodes, exposed via welldefined interfaces and meaningful for the environment. A R C H I M A T E M A D E P R A C T I C A L 2 2

27 Consequences: Not applicable. Alternatives: Not applicable. Relationships with other good practices: Not applicable. A R C H I M A T E M A D E P R A C T I C A L 2 3

28 4.3 Infrastructure landscape Question: How can an infrastructure landscape be modeled in ArchiMate? Enterprise architecture is used to register the essentials of architecture domains and visualize the relationships between those domains. In the technology domain implementation-specific details are often important, but which aspects should and which should not be modeled in an infrastructure landscape in ArchiMate? Solution: When modeling an infrastructure landscape it is essential to maintain distance (or objectivity) from the stakeholder and their corresponding concerns. In most enterprise architecture initiatives the goal of modeling an infrastructure landscape is to visualize the most important elements (hardware and software) of the infrastructure, how these are connected to networks and what their geographical location is (such as main office and regional offices). Limit an ArchiMate infrastructure landscape to the most important physical systems and networks and specific essential supporting software like operating systems, database management systems and middleware. Use the Infrastructure viewpoint from the ArchiMate book (Lankhorst et al., 2005, p. 186). This viewpoint allows the architect to show which essential hardware and software elements determine the infrastructure landscape and how they are connected via networks. It is recommended practice to group infrastructure elements in this viewpoint based on geographical location and make these groups explicit. Figure 4-19: Example of an infrastructure landscape Consequences: Not applicable. Alternatives: Not applicable. Relationships with other good practices: Not applicable. A R C H I M A T E M A D E P R A C T I C A L 2 4

29 4.4 Difference between system software and artifact Question: What is the distinction between the ArchiMate concepts 'System Software' and 'Artifact'? In practice, these concepts are used in different ways. Solution: System Software and Artifact are both concepts from the Infrastructure layer of ArchiMate. System Software is a structural concept, while an artifact is an information concept. The definitions of the two concepts are as follows: System Software: System software represents a software environment for specific types of components and objects that are deployed on it in the form of artifacts Artifact: A physical piece of information that is used or produced in a software development process, or by deployment and operation of a system System Software is a specialization of a node, and is used to model the software environment on which artifacts are deployed. An artifact represents a concrete physical element, and is used to model (software) products, such as source code, executables, scripts, database tables, messages, documents, specifications, etc. Figure 4-20: Relation between system software and artifact Modeling of an element (for example, software) as artifact puts the emphasis on the information-aspect, the physical side (a file, a number of lines of code or a database table). Modeling of an element as system software puts the emphasis on the behavior of that element. There is also a "natural" relationship between artifacts and system software: an artifact is deployed on system software. This is modeled with an assignment relation. Examples would be a database table that is deployed on an SQL Server database, or a DLL on a platform running on Windows XP. Consequences: Not applicable Alternatives: Not applicable Relationships with other good practices: Not applicable A R C H I M A T E M A D E P R A C T I C A L 2 5

30 5 Modeling over the boundaries of layers 5.1 Overview of good practices In this chapter the following good practices are described: 1. Most commonly used relationships. 2. Product and application services. 3. Support for batch and online processes. 4. Service broker. 5. Domain. 6. Internal and external services. 7. Necessity for using services. 8. Simplification and preserving consistency. 9. Grouping 10. Information exchange with external organizations 11. Display messages in queues 12. Nesting and implicit 13. File transfer A R C H I M A T E M A D E P R A C T I C A L 2 6

31 5.2 Most commonly used relationships Question: ArchiMate consists of a considerable number of concepts and relationships. The question of which relationship is used to model between two concepts is exactly defined in ArchiMate. In day-to-day practice it can be hard to pick the right choice from the allowed possibilities. The question must focus on which relationships between concepts should be used in the most practical cases. In the section below we describe which relationships should be modeled for these cases. To prevent misunderstandings: in the list summarized below not all ArchiMate relationships between concepts are described, only those relationships that we considered the most often used. A complete overview of all possible relationships between different concepts can be found in the ArchiMate reference manual. Solution: Below an overview of most commonly used relationships for the concepts: Business service Business process Business actor Business role Business object Business product Application component Application service In particular those related to the business layer and application layer. Business service: a business service is realized by a business process or a business function; a business service is used by a business role; a business service has access to a business object (a business service creates, reads, modifies or destroys a business object); a business service can consist of other business services and can use other business services. Figure 5-21: Business service - most commonly used relationships between concepts A R C H I M A T E M A D E P R A C T I C A L 2 7

32 Business process: a business process realizes a business service; a business process exchanges data with other business processes (via the flow relationship); a business process is triggered by or triggers a business event, a business function or other business processes; a business process is assigned to a business role; a business process is part of a business function; a business process has access to a business object (a business process creates, reads, modifies or destroys a business object); a business process uses application services. Figure 5-22: Business process - most commonly used relationships between concepts Business actor: A business role can be assigned to a business actor. Figure 5-23: Business actor - most commonly used relationship between concepts A R C H I M A T E M A D E P R A C T I C A L 2 8

33 Business role: A business role can be assigned to a business actor; A business role can be assigned to a business process, business function or business event; A business role can use a business interface. Figure 5-24: Business role - most commonly used relationships between concepts Business object: A business object is created, read, modified or destroyed by a business process or business function (via access relationship); A business object can have specializations; A business representation realizes a business object; A business object can point to other objects (aggregation relationship); A business object can consist of other objects (composite relationship). Figure 5-25: Business object - most commonly used relationships between concepts A R C H I M A T E M A D E P R A C T I C A L 2 9

34 Business product: A business product consists of services and contract(s); A business product represents a value. Figure 5-26: Business product - most commonly used relationships between concepts Application component: An application component realizes an application service; An application component has one or more application interfaces; A data object is created, read, modified or destroyed by an application component or application function; An application component can be part of an application collaboration (via aggregation relationship); An application component can consist of multiple application components (via composite relationship); An application component can be assigned to an application function; An application component uses infrastructure services; Data flows can exist between application components (flow relationship). Figure 5-27: Application component - most commonly used relationships between concepts A R C H I M A T E M A D E P R A C T I C A L 3 0

35 Application service: an application service is realized by an application component or an application function; an application service is used by an application component; an application service has access to a data object (an application service creates, reads, modifies or destroys a data object); an application service can consist of other application services and can use other application services. Figure 5-28: Application service - most commonly used relationships between concepts Data object: A data object is created, read, modified or destroyed by an application component, application service or application function (via access relationship); A data object can have specializations; An artifact realizes a data object; A data object can point to other data objects (aggregation relationship); A data object can consist of other data objects (composite relationship). Figure 5-29: Data object - most commonly used relationships between concepts A R C H I M A T E M A D E P R A C T I C A L 3 1

36 Consequences: Beware; these are a number of frequently occurring situations where the use of relationships in ArchiMate is much broader than summarized above. A complete overview can be found in the Architecture Language Reference Manual. Alternatives: Not applicable. Relationships with other good practices: Not applicable. A R C H I M A T E M A D E P R A C T I C A L 3 2

37 5.3 Product and application services Question: Isn t it strange that a product can not only consist of business services but also application- and infrastructure services? A customer cannot possibly consume application services? So if an application service is used almost directly by a customer, wouldn t there be a business service in between? In some examples in the ArchiMate Specification, a product is composed of application and business services. Solution: The Architecture Language Reference Manual indeed contains examples where customers use application services directly. It is better to put a business service in between. In the product description of the Architecture Language Reference Manual (p. 26) the application services Account status and Money transfer will become business services. The connection between a business service and an application service is routed via a business process that supports the business service, and is in its turn supported by an application service. This validates the indirect relationship that a business service uses an application service. Consequences: You should consider deleting the aggregation relationship between product and application service. Alternatives: Not applicable. Relationships with other good practices: Not applicable. A R C H I M A T E M A D E P R A C T I C A L 3 3

38 5.4 Support of batch and online processes Question: How can you model two variants of the same process in ArchiMate whereby one process is executed as a batch process and the other process is executed manually (online)? Solution: Distinguish two variants of the business process: one for manual (online) processing and one for batch processing. Although the results of the processes can be identical, the manual (online) processing allows more room for exceptions and often more granular process steps can be distinguished. In both process descriptions it is sensible to consider actions as units of time, place and behavior when determining the size of the actions. This will probably differentiate the two process variants. Based on the two variants of the business process, describe the application services that support these processes. Use the actions that have been modeled in the business process as a starting point: the implication of this is that the application services for both variants can differ! Often it will be that application services that support a batch process are more granular than application services that support an online process, because in the latter case more granular actions can be distinguished. Beware, it is very well possible that application functions that realize the different functions of application services are common: this points to reuse of functionality in the batch and online processing. Use the Application usage viewpoint from the ArchiMate book (Lankhorst et al., 2005, p. 184). This viewpoint allows one to visualize how business processes are supported by applications, via application services. Figure 5-30: Example of a batch and online (manual) variant of the same process Consequences: Not applicable. Alternatives: Not applicable. Relationships with other good practices: Refer to the good practices Business service, business function and business process, Constraining business processes and Viewpoints for process modeling. A R C H I M A T E M A D E P R A C T I C A L 3 4

39 5.5 Service broker Question: How do you model a service broker (or comparable component, for example a workflow handler or a middleware-type component)? On one hand it is clear that the broker is the center point of all services, but on the other hand it is not visible which application produces which service (via the broker). The goal of a broker is simplification of the many relationships between different applications. Sometimes the following modeling pattern is used: Figure 5-31: Service broker related to applications - undesired A disadvantage of this way of modeling is that it is no longer evident which application delivers which service. It is also desirable to hide or unhide the broker (and its service) when required. This is not possible with the above modeling pattern. Solution: The core of the solution is in relating the applications and application services on one hand (the application realizes the corresponding application service), and relate the application services and the broker service on the other hand (the broker service aggregates the application services). The situation described above will then be modeled as follows: Figure 5-32: Service broker related to applications - desired A R C H I M A T E M A D E P R A C T I C A L 3 5

40 By generating different views (and leaving out certain relationships or showing these in a nested manner), different focal points can be achieved. In the following figures the relationship between applications with and without a broker and the role of the broker are visualized. Figure 5-33: Service broker related to applications - views Consequences: By modeling a broker and a broker service as shown above, it becomes possible to either show or hide the broker at will, which will improve the insight of the model for different stakeholder groups. Alternatives: Not applicable Relationships with other good practices: The service broker is part of the application landscape. Refer to good practice Application landscape. A R C H I M A T E M A D E P R A C T I C A L 3 6

41 5.6 Domain Question: How can I model a domain in ArchiMate? A domain is defined as a certain demarcation meant to structure concepts. These could be business-, application-, information- or technical domains. (Note that the term domain is used within TOGAF in a much stricter sense.) Solution: In ArchiMate the grouping relationship is suited to grouping domains. The relationship can be given a name. In the example below (the part of) the sales domain is shown. Figure 5-34: Example of a sales domain Consequences: Not applicable Relationships with other good practices: Refer to good practice application landscape. A R C H I M A T E M A D E P R A C T I C A L 3 7

42 5.7 Internal and external services Question: ArchiMate distinguishes services that are used within the same layer in which they are situated and by the next layer above. The same layer is the business layer for business services, the application layer for application services and the infrastructure layer for infrastructure services. ArchiMate calls this type internal services. By contrast those services used by elements of the next layer above the layer in which they are situated are referred to as external services The next layer above the infrastructure layer is the application layer. This is the layer in which external infrastructure services are used. Similarly, external application services are used in the business layer and external business services are used by the external environment (e.g. by customers). The concepts internal and external are therefore relative with respect to the layer where the service is being realized. How do you visualize this difference? Also refer to EA at work, p. 86. Solution: Defining services that are used within their own layer ( internal services ) is what is often meant with Service Orientation. The modeled services usually coincide with actually implemented / to be implemented services (especially with respect to application services and infrastructure services). Services that are used by the layer above ( external services ) represent meaningful functionality that supports the layer in which they are are used and will not always refer to an implemented service (with respect to an actual externally callable functionality). Our recommended good practice is to handle internal and external services as two dedicated groups and distinguish these in views, especially in those cases where distinction between internal and external services is important. External services will often have other demands than internal services, and other aspects will have to be registered. The service itself will often have a differing design, depending on the intended support within the own layer or the next upper layer. Figure 5-35 is an example of how this distinction can be visualized: the internal and external services are distinguished by placing (grouping) them in separate layers 1. In this example the Customer management service is an external service and is therefore positioned in the corresponding layer. The Customer data service is used only within the application layer and is therefore an internal service. Another possibility is to shape and/or color internal and external services accordingly. 1 Notice that in Figure 5-35 a simplification has been applied by leaving out the application function refer to good practice "Simplification with sustainable consistency". A R C H I M A T E M A D E P R A C T I C A L 3 8

43 Figure 5-35: Internal and external services Consequences: Make an explicit distinction between internal and external services. Alternatives: In this good practice a distinction is made between internal and external services. Reasons may exist not to make this distinction but to highlight another distinction instead, for example between business critical and non critical services, or between services with a gold, silver or bronze service level agreement. To make a distinction like this an identical approach to that described above can be used, e.g. to use grouping, coloring or shaping to emphasize the distinction. The model and visualization goal determines the distinction that needs to be emphasized. Relationships with other good practices: There is a relationship with good practice Interfaces between applications. A R C H I M A T E M A D E P R A C T I C A L 3 9

44 5.8 Necessity for the usage of services Question: Are services mandatory in ArchiMate? In my current situation we have no implemented services. We also do not have a service-oriented vision. Solution: Services in ArchiMate are not mandatory. If these are not distinguished and not foreseen for the future, the services can be left out at will. Consequences: If no services are distinguished, then no services will be modeled. The rules applying to good practice Simplification with sustainable consistency have to be used to use the ArchiMate concepts in a consistent manner. Instead of a process that uses a service (1), this uses an application function (2) or, with increasing simplification, an application (3). 1) 2) 3) Figure 5-36: Use of an application function instead of an application service Alternatives: Even if currently no services are distinguished, service-oriented thinking can help to achieve another view on the world. This can lead to new insights. By considering the world from a service-oriented viewpoint one often will arrive at other groupings and responsibilities. If the service-oriented approach is envisioned to be used in the future, that future situation can be modeled with services. Relationships with other good practices: Refer to good practice Simplification with sustainable consistency. A R C H I M A T E M A D E P R A C T I C A L 4 0

45 5.9 Simplification and preserving consistency Question: How can I simplify my views without compromising the consistency? Solution: Within ArchiMate it is possible to deduce indirect relationships by compositions of relationships. This is described in paragraph 5.7 of the book EA at work. This is considered an indispensable instrument for visualizing architecture: leave out some details and at the same time maintain consistency between models. For structural relationships, a powerful composition rule has been developed based on the strength of the relationship. In a chain of concepts the most weak relationship in the chain determines the overall relationship between the endpoint concepts of the chain. The table below shows the strength of structural relationships. Figure 5-37 shows an example of this composite rule. In this example, the indirect relationship between the process Registration and the application function Manage Customer data is a used by relationship because this has a lower weight than a realization. Table Strength of structural relationships Relationship Strength association 1 access 2 used by 3 realization 4 assignment 5 aggregation 6 composition 7 Figure 5-37: Example of an indirect relationship used by The following rules apply for the two dynamic relationships: A triggering or flow relationship between behavioral elements (for example processes or functions) may be redirected to structural elements (for example business actors or application components) to which they are assigned. A R C H I M A T E M A D E P R A C T I C A L 4 1

46 Figure 5-38: Example of an indirect relationship - flow A triggering- or flow relationship between behavioral elements may be redirected to the services that they realize. Figure 5-39: Example of an indirect relationship - trigger Consequences: Views can be simplified in this manner to a level of detail that is relevant for the goal of the view, without losing consistency. Not every tool supports this simplification. The consequences are that redundant relationships may exist. In the above example of the services this implies that the existing relationship between the processes is registered in the tool (left part). If the relationship between the services is visualized in another view (right part), the tool may introduce an extra triggering relationship between these services. Alternatives: Not applicable Relationships with other good practices: These simplifications can be applied to any good practice. A R C H I M A T E M A D E P R A C T I C A L 4 2

47 5.10 Grouping Question: What mechanisms are available in ArchiMate to represent grouping? Solution: ArchiMate has a separate concept for grouping (actually it is a relation to represent grouping), as shown below. This grouping concept allows for the (visual) grouping of an arbitrary collection of objects. Figure 5-40: Grouping concept This visual grouping does not have not enough semantic power, however, for many situations. Thus, ArchiMate also supports various other mechanisms for semantic grouping. Almost all concepts can be hierarchically grouped using composition or aggregation relations. This technique allows the modeler to indicate that a process consists of sub processes for instance, or that an organization consists of departments. This mechanism is mostly used to group similar concepts. Figure 5-41: Grouping elements of similar concepts Using these relations, different types of concepts can be grouped together. An example is a business product that aggregates a contract and several business and application services. Figure 5-42: Grouping elements of different concepts A R C H I M A T E M A D E P R A C T I C A L 4 3

48 ArchiMate also supports grouping arbitrary concepts using the association relation. An example is the modeling of ownership or responsibility, where a certain role is the owner or responsible entity for a number of applications, processes, and business functions. These concepts are connected using the association relation with that role (the relationship can be categorized as "ownership" in that case). Figure 5-43: Grouping using association relation Using this mechanism, arbitrary groupings can be created and visualized. In many cases also the term "domain" is used to denote grouping. Examples are the domain "Customer", "Life", etc. Often the domain can be modeled using a concept like business function, business role, or application function, where the other elements are grouped within that concept. Not every grouping needs to be modeled explicitly. In many cases, a group is nothing more than a view from a certain perspective. For instance, an overview of all processes supported by applications owned by one specific role. This grouping can be determined by the relations from that role via the applications to the processes. Consequences: Not applicable. Alternatives: One of the possible extensions of the ArchiMate language is a generic grouping concept; extending the language with such a concept will of course solve the above problem. Using supporting tools it is in a number of cases possible to extend the meta model of the tool with new concepts and relationships. In that case it is possible to add a new grouping concept, additional to the ArchiMate language. If the ArchiMate language is extended with a grouping concept, a conversion of the models is of course necessary. Relationships with other good practices: This good practice is related to the good practice domain A R C H I M A T E M A D E P R A C T I C A L 4 4

49 5.11 Information exchange with external organizations Question: How you model data exchange with external organizations on the application layer? At the application level, there is indeed application functionality that realizes these information exchanges. But should you also include the external application in the application landscape? Solution: A first alternative is to model a flow relation directly from the native application to the external organization. Here the external application is not included. Figure 5-44: Flow relation between internal application and external organization This prevents us from including external applications in the application landscape. For data exchanges with various external organizations it would increase the number of external applications in the application landscape, which could become cluttered as a result. See the example below: Figure 5-45: Flow relation between its own application and the external organization application A second alternative is to model an external application service. Then the data exchange between application services is modeled. A R C H I M A T E M A D E P R A C T I C A L 4 5

50 Figure 5-46: : Flow relation between the application services of its own application and the external organization If desired the flow relations between the application services can be replaced by a dataobject; written by one application service, read by the other application service Consequences: Not applicable. Alternatives: Not applicable. Relationships with other good practices: Here is a relationship with the good practice 'Coupling between applications' A R C H I M A T E M A D E P R A C T I C A L 4 6

51 5.12 Display messages in queues Question: Messages that are received are processed in steps. Each step will change the status of the message. There is a separate queue for each message status in MQ-series, enabling further processing by the same or a different application. Insight is necessary regarding in which queue a message with a certain status is stored in order to be processed by what application. How do you model that? Solution: The solution given here provides a pattern for the above question. It is not an explicit example. The core of the solution is that each queue is modeled as an artifact in the infrastructure layer. These queues are then connected to the queue manager with an assignment relation. The queue manager is system software that is running on a node. The message with the appropriate status is also an artifact, and is connected to its queue an aggregation relationship. These artifacts realize data-objects in the application layer. There may be as many queues, messages and data objects modeled as necessary. In the example below there are two, but that can be expanded with any desired number. Queues will typically have more technical names than the names used in the example. Figure 5-47: Distinct messages in different queues Consequences: If many queues and messages are modeled in this way, the model can become confusing. Also, the message is drawn several times in the application layer, which may suggest that there are several different data-objects. That however is not the case. It is the same logical data-object, but each time with a different status. Physically they are indeed separate XML messages. A R C H I M A T E M A D E P R A C T I C A L 4 7

52 Alternatives: When it is important to stress in the model that a message represents the same data-object, there are two alternatives. The first alternative shows the data-object just once. The specific status of the message is then reflected in the access-relationship between the data object and the application. This relationship then models the status. Figure 5-48: Describe the status of message in the access relation Another possibility is to model a message using specializations for the different statuses of the original message. This indicates that on the one hand this is still the same data-object, but on the other hand can be used to explicitly show the different states. The specializations of the data object can be linked to applications. The figure below shows this way of modeling. Figure 5-49: Describe the status of message via specialisation of data-object Relationships with other good practices: Refer to good practice 'Difference between system software and artifact'. A R C H I M A T E M A D E P R A C T I C A L 4 8

53 5.13 Nesting and implicit relations Question: When nesting concepts, do the relationships of the nested object also implicitly apply for the 'parent' object? Solution: The concept of nesting is not formally defined in the ArchiMate language. In practice however various forms of nesting occurs 1. A visual representation of nesting different concepts. Figure 5-50: Mix of nested concepts The advantage of this form is the relatively simple view where relationships are abstracted between different objects towards there is a relationship. The disadvantage is that in this view it is not known what the exact relationships between the objects are. 2. Hierarchical decomposition: This form of nesting is often used to decompose the same concept into sub concepts. Examples are hierarchical process, function, or organizational structure. The following example shows a process hierarchy. A R C H I M A T E M A D E P R A C T I C A L 4 9

54 Figure 5-51: Nesting through a process hierarchy Variant 1 indicates the visually nested sub-processes. The relationships between the sub processes and the main process are not visible (but should formally be modeled). Variant 2 indicates, in this case, the composite relationships between objects. When nesting between the same concepts it is common to use the composition or aggregation relationship. Both variants model Subprocess-3 with an Access relation towards Object-1. Subprocess-3 is decomposed from Main Process. According to the rules of derived relations (see good practice 'simplifications and preservation consistency') it holds true that the main process also has the Access relation has with Object-1. Hence with this type of nesting the parent object has the same relation as the child object. Note: Composition or decomposition of a concept is not the same as clustering of similar concepts. For the latter, the concept of 'Grouping' may be used Figure 5-52: Clustering using grouping Consequences: When visual nesting of objects the relationships between these objects must be explicitly described. Alternatives: Not Applicable. Relationships with other good practices: See also good practice: Simplification and preserving consistency. A R C H I M A T E M A D E P R A C T I C A L 5 0

55 5.14 File transfer Question: How to model file transfer, i.e. FTP? Solution: There are several ways to visualize file transfer using ArchiMate. The key is to use the technology usage viewpoint as a basis. This viewpoint describes the infrastructure services used by applications. File transfer can be modelled as an infrastructure service used by two applications: the provider and consumer. The realisation of the file transfer infrastructure service requires both the provider as the consumer to be deployed with supporting system software. Figure 5-53: File transfer From the above view we can see the application interface Trans info between ACME application X and Y. ACME Application Y is the provider of the interface and ACME Application X uses Trans Info interface. This is depicted by the composition and used by relationship. The Trans info application interface uses the Internal File transfer infrastructure service. The internal file transfer infrastructure service is realized by system software a) Coyote Slow FTP Client - deployed in same environment as ACME Application X and; b) RoadRunner Quick FTP Server - deployed in same environment as ACME Application Y In a technology usage view it serves no purpose to indicate the flow of data. It can however be concluded that the flow is from X to Y. A flow from Y to X would require an additional application interface. Consequences: Usage of this good practice requires the usage of application-interfaces as a substitute or addition to flow relations. Alternatives: Not applicable. Relationships with other good practices: There is a relationship with the The Display messages in queues good practice. This good practice may also be used for infrastructure services like Messaging. In that case the system software would be based Message Queue software. A R C H I M A T E M A D E P R A C T I C A L 5 1