A Con gurator for Visual Analysis of Enterprise Architectures

Abstract

Enterprise Architecture (EA) management seeks to mutually align business and IT while fostering exibility to react upon environment changes appropriately. Highly distributed data within the entire enterprise is collected to facilitate decision making processes during enterprise transformations. At the same time enterprise architects and EA stakeholders have new and arising questions hard to predict in advance. Visualizations are a common means for decision makers to analyze complex information about the entire enterprise. Thereby, these visualizations are commonly generated using model-driven approaches. Although, a common set of best-practice EA visualization types could be distilled, their actual binding to underlying data and organization-speci c con guration is still challenging for business users. In order to cope with this challenge, this demo1 paper presents a con gurator for EA visualizations end-users without an IT background are able to create spontaneously.

Keywords

• Enterprise Architecture
• visualization
• collaboration

-

the actual EA management-problem to be addressed, as the key stakeholders of the EA, namely enterprise architects and C-level executives, require di erent information during decision making. Additionally, the relevant information has to be presented in a stakeholder-speci c way in graphical architecture descriptions [ 7 ]. The employed architecture views vary widely with respect to the used visualization elements, but also with respect to the information visualized.

Software

engineer

Data custodian

Data steward

Designer

End-user

Repository

d p o r a

Modeling

EA Visualization

Databinding

Visualization Design

Configuration

Rendering

Scope of this paper

Required periphery

Communication effort

In recent years, an approach to support the automated generation of graphical EA descriptions based on arbitrary underlying information models has been developed [ 8, 7, 2 ]. Central to this approach is the idea to describe a viewpoint as a model-to-model transformation from the information model to the visualization model [ 1 ], an expressive model for describing visualizations. A particular transformation de nes which information model concepts are translated into which visualization elements. Research in close cooperation with practitioners has shown that this approach can be bene cially applied in the context of EA management. Nevertheless, the development and de nition of the actual modelto-model transformation remains a central but complex part of the design process, as prevalent model-transformation languages are complex to understand, especially for the stakeholder of an EA visualization tool. Further, the application cases revealed that users want to perform customizations, i.e. minor adaptations, of the visualizations with respect to the visualized information, e.g. by replacing one type from the information model with a di erent one. A possible example would be `database management systems per server' instead of `business applications per server'. The delineated shortcomings for visualizations in the context of EA management motivate the objective addressed in this demo paper: How can business users be empowered to con gure visualizations that are bind to an arbitrary EA information model?

Con gurable

Enterprise

Architecture

Visualizations

Prevailing approaches for generating EA visualizations typically take into account multiple steps p erformed by a variety of di erent actors as displayed in Figure 1. During the mo deling step the required information is gathered from stakeholders or existing information sources in the organization (cf. e.g. [ 5 ]). The result of the data binding step is a mapping, i.e. mo del-to-mo del transformation, b etween information mo del elements and an abstract viewp oint de nition, i.e. a visualization typ e. As motivated ab ove, typically this step requires sp eci c knowledge from exp ert users. The de nition of the abstract viewp oint is p erformed within a visualization design step. During con guration, the abstract viewp oint is detailed with e.g. color enco dings, axis descriptions, and symb ols. As a result, these subsequent steps lead to a communication e ort b etween the various actors. We aim at providing b etter end-user supp ort for the creation of these con gurable EA visualizations. In particular, we fo cus on the data binding and con guration to reduce the required communication e ort.

1.  S elect  v isualiza/on   EA  (cur ent)   Ap licaHon  Service  (50)   Busines  Object  (15)   Data  Object  (150)   Ap icNaHoodne C Doemvipceon (5e0n9t ) ( 28) 4 .  Data Ap licaHon  Interface  (50)   Ap icaHon  Component  (28)   AIpnfricastHrounct Cuorem Speornveicnet (1258)   Ap licaHon  Component  (28)  

PStaIrCddno PdluraaesCcrcaesttusoullttaacair adroerobbiaeebmnumraancmadtcsscoes t eeee rre dn        CCluuCCssumtutesoseOrtmro SAX eA1 rP4    CRPMPZr1O o0tnd  3luin4ce    SSCAauPPNltesr  EAPotsoRFlNdamoPuOnrc e4c   r.e1      SaNMMleeaospaFtsTleoe urrSnic  e    CProCiSaoamcnLttoe rssa  21   

Sales   MaFin  

Mercur    Binding

 

6.  Visual  C

onfigura/on  

8.  Visual  Manipula/on  

Screenshots of the demo and high-level architecture

3.  Matching  r

JSON   5.  F ilt ering   7.  F inalizing  

VisualizaHon   Type       InformaHon   Demand   EA  

Repository   esults   InteracHve   VisualizaHon   Framework   Enterprise  2.0 Framework     RaphaëlJS   Customer  cord  Client da b se  Insurance  laim  Product  a b se  Procu em nt da b se  Sales r co d 

Customer  CCluusstteorm Xe1r4  SAP CRM  CusOA 

Product  Z10 34 P Onli e 

Product Plan Customer  SAP ER 4.1  SalesForce  NA O 

Materi l  SalesForce  MoseS  NepTun 

Contrac   Priamos 2 CoSaLes 1

Sales  MaFin  Mercu   2.  S tructural  P a5ern  

Matching  

Inform atio n

Offer Pow erp oint   2.1

Demonstration We employ structural pattern matching of models [ 6, 2 ] to recommend feasible model-to-model transformations based on a chosen visualization type and given EA information model. The former speci es the demand model whereas the latter requires the o er model. In a rst step, an end-user may choose from a list of o ered best practice visualization types that have been identi ed in previous projects (cf. Figure 2). These visualization types are queried from the backend-system. Based on the chosen visualization type, the information demand is determined automatically. In the next step, the data binding is speci ed by the user. Thereto, the business users does not have to fully grasp the entire EA information model and can concentrate on the analysis of the EA. In order to recommend feasible con gurations for a data binding, the information demand of a visualization type and information o er from the EA repository's information model are matched structurally. The outcome of this step de nes a model-to-model transformation. Especially transitive relationships are also autorecommended which also lowers the barrier for more complex queries. Selected elements of an EA information model can be ltered subsequently by the enduser. This lter is applied to the instances, e.g. to use all business applications with criticality high. Next to this step, the con guration of the visual design can be speci ed in detail to customize the visualization. These visual parameters are precon gured variability points of a particular visualization type, e.g. size, shape, or color used within a visualization. The visualization is then inserted on an ordinary wiki page and may be annotated or viewed within the content area of a wiki page that is stored in the EA repository. For executive presentations, any visualization on a wiki page can also be downloaded in fully compatible PowerPoint (pptx) format. This way, users can manipulate generated visualizations even beyond the model-driven approach, i.e. just edit colors for reasons not visible in the actual data displayed as it is commonly the case for tacid knowledge required for a particular management meeting. 2.2

Implementation We utilized an existing Enterprise 2.0 framework [ 3 ] which allows us to manage structured information using a web browser or import data from arbitrary information sources in the organization. This system is developed in Java and has already been applied as a lightweight EA repository in practice. We extended this framework with interactive visualizations capabilities that build upon the concepts presented in [ 7 ]. The actual visualization rendering within the browser is accomplished using the RaphaelJS2 framework to provide cross-browser support. The pattern matching between the abstract viewpoint and the information stored in the Enterprise 2.0 framework is achieved using the algorithm proposed in [ 6 ]. Changes to these generated visualizations are propagated to the EA repository such that underlying data is manipulated. 2 http://raphaeljs.com/, last accessed on: 2013-07-15

Conclusion

In this paper we presented our demonstration system that empowers business users to con gure visualizations that are bound to an arbitrary EA information model. Thereby, the major design goal was to lower the barrier for stakeholders with a non-technical background analyzing the EA. Special to our approach is the recommendation of feasible con gurations to the end-user such that it is not required to fully understand the entire EA information model. Our solution supports EA stakeholders that want to analyze the EA and derive plannedstates in a rapid manner without manipulating underlying data. We o er the capability to export the visualization as an editable PowerPoint presentation. In future work we plan to identify further abstract viewpoints from practice and extend the con gurator with new capability for the analysis of EA models.

Acknowledgment

This research has been sponsored in part by the German Federal Ministry of Education and Research (BMBF) with grant number TUM: 01IS12057.

References

1. Alexander M. Ernst , Josef Lankes, Christian M. Schweda , and Andre Wittenburg . Using model transformation for generating visualizations from repository contents { an application to software cartography . Technical report , Technische Universitat Munchen, Chair for Informatics 19 ( sebis ), Munich, Germany, 2006 .

2. M. Hauder , F. Matthes , S. Roth , and C. Schulz . Generating dynamic crossorganizational process visualizations through abstract view model pattern matching . In Architecture Modeling for Future Internet enabled Enterprise (AMFInE) , 2012 .

3. Florian Matthes , Christian Neubert , and Alexander Steinho . Hybrid wikis: Empowering users to collaboratively structure information . In Mar a Jose Escalona Cuaresma , Boris Shishkov, and Jose Cordeiro, editors, ICSOFT (1) , pages 250 { 259 . SciTePress , 2011 .

4. Jeanne W. Ross , Peter Weill , and David C. Robertson . Enterprise Architecture as Strategy . Harvard Business School Press, Boston, MA, USA, 2006 .

5. Sascha Roth , Matheus Hauder, Matthias Farwick, Ruth Breu, and Florian Matthes . Enterprise architecture documentation: Current practices and future directions . In 11th International Conference on Wirtschaftsinformatik (WI) , Leipzig, Germany, 2013 .

6. Sascha Roth , Matheus Hauder, Marin Zec, Alexej Utz, and Florian Matthes . Empowering business users to analyze enterprise architectures: Structural model matching to con gure visualizations . In Trends in Enterprise Architecture Research (TEAR) , Vancouver, Canada, 2013 .

7. M. Schaub , F. Matthes , and S. Roth . Towards a conceptual framework for interactive enterprise architecture management visualizations . In Modellierung, Bamberg, Germany, 2012 .

8. Andre Wittenburg . Softwarekartographie: Modelle und Methoden zur systematischen Visualisierung von Anwendungslandschaften . PhD thesis , Fakultat fur Informatik, Technische Universitat Munchen, Germany, 2007 .