Conceptualizing Data in Multinational Enterprises - Springer Link

3 downloads 161811 Views 164KB Size Report
Therefore data management has to provide enterprise-wide data ownership, ... a conceptual model, which covers all requirements for defining, governing, us- ... concepts enabling multinational enterprises to meet business requirements. .... refers to physical data storage, being centrally in one application or distributed over.
Conceptualizing Data in Multinational Enterprises: Model Design and Application Verena Ebner *, Boris Otto, and Hubert Österle Institute of Information Management, University of St. Gallen, St. Gallen, Switzerland {verena.ebner,boris.otto,hubert.oesterle}@unisg.ch

Abstract. Collaboration and coordination within multinational enterprises need unambiguous semantics of data across business units, legal contexts, cultures etc. Therefore data management has to provide enterprise-wide data ownership, an unambiguous distinction between "global" and "local" data, business-driven data quality specifications, and data consistency across multiple applications. Data architecture design aims at addressing these challenges. Particularly multinational enterprises, however, encounter difficulties in identifying, describing and designing the complex set of data architectural dimensions. The paper responds to the research question of what concepts need to be involved to support comprehensive data architecture design in multinational enterprises. It develops a conceptual model, which covers all requirements for defining, governing, using, and storing data. The conceptual model is applied in a case study conducted at a multinational corporation. Well-grounded in the existing body of knowledge, the paper contributes by identifying, describing, and aggregating a set of concepts enabling multinational enterprises to meet business requirements. Keywords: Enterprise data architecture design, enterprise data architecture management, data quality management, data modeling, data classification.

1

Introduction

Consumer-centric business models pose the demand for a 360° perspective of the customer, increasing value chain integration creates the need for business collaboration and information sharing [1, 2], and a fast-growing number of legal regulations and contractual obligations requests consolidated and integrated data across the enterprise. Especially in multinational enterprises, these needs for enterprise-wide collaboration, coordination and interoperability are faced by ambiguous definitions of enterprise data1 across multiple business units, legal contexts, as well as geographical regions, numerous stakeholders and missing responsibilities for enterprise data, multiple distributed, heterogeneous, internal and external applications storing and managing enterprise data in a redundant and inconsistent manner, and a variety of business processes using and managing enterprise data with different goals. 1

Enterprise data refers to what in literature is commonly named master data. To distinguish from the definitions of master data in the practitioners’ community, the term enterprise data is used here.

P. Atzeni, D. Cheung, and R. Sudha (Eds.): ER 2012, LNCS 7532, pp. 531–536, 2012. © Springer-Verlag Berlin Heidelberg 2012

532

V. Ebner, B. Otto, and H. Österle

Enterprise data in multinational enterprises can be described with regard to the characteristics time reference, change frequency, volume volatility and existential independence [3]. Enterprise data stores and describes characteristics of a company’s core business entities, e.g. customers, suppliers, or products [1]. Shared access, replication, and flow of enterprise data in order to ensure data quality is controlled in the enterprise data architecture [4]. Enterprise data architecture aims to support collaborative use and management of enterprise data by providing an enterprise data model, enterprise data applications and a description of data flow between applications [3, 5]. Therefore, it is necessary to assess, describe and document business requirements to be met by enterprise data on an attribute level. The paper takes up on the research question as to what are the enterprise data architecture design decisions and which criteria, also called classifiers, best support the assessment of alternative design options. More precisely, the paper aims to answer the question as to what are the components a conceptual model for enterprise data needs to involve, so that all criteria to define, govern, store and use data in the environment of a multinational enterprise are taken into consideration. A conceptual model is designed, that aims to identify, describe and aggregate a complex set of classifiers enabling multinational enterprises to design an enterprise data architecture that meets business requirements of pressing relevance. The conceptual model was designed by the analysis of existing conceptual data models, and a literature review identifying enterprise data conceptualizations. The results were reflected in two multiple expert interviews. Subsequently, the applicability of the model was demonstrated in a realworld context at a multinational electronic and electrical engineering corporation [6].

2

Model Design

The conceptual model for enterprise data shown in Fig. 1 contains both structural and behavioral components [7]. The structural components reflect the enterprise data elements, e.g. classes, attributes, or their relationships. The behavioral component describes the enterprise data elements. The behavior may differ for each enterprise data element. An element can be described from four different perspectives, so called views, namely Administration, Governance, Storage and Usage [8]. Administration is concerned with the definition, description and instantiation of an enterprise data element [9]. Governance focuses on data ownership and data management processes. Storage handles the distribution of data between applications. Usage refers to the use of data in business processes [10]. For each view, a set of characteristics is specified called enterprise data classifiers which can be operationalized by value sets. Tab. 1 shows a reference list of enterprise data classifiers. The list is exemplary and by no means complete. The Administration view supports an unambiguous definition of enterprise data within the enterprise with regard to its meaning and semantics [1]. Validity refers to the reach of the definition to be compulsory within the organization. It can be valid for the whole enterprise (global), for some parts of the enterprises (inter-divisional), or in one single enterprise division only (local). Data definition autonomy describes

Conceptualizing Data in Multinational Enterprises: Model Design and Application

533

how independent data can be defined among various business units. For some enterprise data elements, e.g. unique identifiers, a distinct instantiation of data values is necessary [2], i.e. Uniqueness. The concept of Structural constraints describes structural guidelines. There can be no structure given at all, some elements of the structure, or the complete structure can be given. Data volume denotes the number of entities handled for the data element. Further classifiers are Change frequency, i.e. the occurrence of modifications of data values, Versioning, i.e. the ability to track changes, and Historiography, i.e. archiving of values. The two latter classifiers are often derived from regulatory specifications.

Fig. 1. Conceptual model for enterprise data

The Governance view relates to organizational aspects of managing data instances. Governance classifiers refer to organizational roles, i.e. Data ownership, and Data stewardship [11]. While the former role is responsible for the definition, the latter is responsible for the creation and instantiation of an enterprise data element. A distinction is made between the scope and the distribution of these roles. While scope defines the area of responsibility and accountability, distribution defines the location, which can be central, in one division, or distributed over multiple divisions [12]. Classifiers concerning data stored in applications and data exchange among applications are enclosed in the Storage view. Data storage applications distinguish the dimensions distribution, autonomy and heterogeneity [13]. Application distribution refers to physical data storage, being centrally in one application or distributed over many applications. Application autonomy denotes the ability of applications to independently process data. With regard to heterogeneity it can be distinguished between Application heterogeneity, and Data model heterogeneity [13]. Application integration concerns the physical merging of data into one application, or the logical mapping of data from distributed applications. Redundancy relates to data being stored in multiple applications. Thereby, identical copies of the same instance (duplicates), different attributes (consolidation), or different instances of the same class (aggregation) can be distributed among multiple applications. With regard to the exchange of data between applications and provisioning to data consumers the following classifiers were identified: Application accessibility, Data provision, Processing, Distribution type, Distribution direction, and Distribution initiation [14].

534

V. Ebner, B. Otto, and H. Österle

In the Usage view focuses on data in business processes. As enterprise data quality is often described as “fitness for use” [15], enterprise architecture design highly depends on the usage of data in business processes. It is distinguished between core and management processes, and support processes [16]. Data Usage types can be analytical or transactional scenarios [2]. Process diversity provides an indicator for data being used differently in various processes. Process dependency refers to multiple processes that can be interrelated. They may depend on one another along the value chain (horizontal) or between divisions (vertical). The Transaction rate provides information about the frequency in which data is exchanged between processes. Table 1. Enterprise data classifiers Administration Governance - Data definition validity - Data ownership scope - Data definition autonomy - Data ownership distribution - Uniqueness - Data stewardship scope - Structural constraints - Data stewardship - Data volume distribution - Change frequency - Versioning - Historiography

3

Storage - Data distribution - Application autonomy - Application heterogeneity - Data model heterogeneity - Application integration - Redundancy - Application accessibility - Data provision - Processing - Distribution type - Distribution direction - Distribution initiation

Usage - Usage type - Process heterogeneity - Process dependency - Process types - Transaction rate

Model Application

The conceptual model was applied at a multinational electronic and electrical engineering enterprise (hereinafter called EEE). When setting up an enterprise wide data management organization, a framework for managing enterprise data was created and an enterprise wide set of goals was specified. Country specific regulations and enterprise wide consolidation of enterprise supplier data to gain strategic benefits for negotiations with vendors and for external e-business processes resulted in the need for world-wide transparency, consolidation, clear and unambiguous responsibilities on an enterprise wide level and standardized processes to maintain and ensure the quality of enterprise supplier data. To identify similarities and differences in handling supplier data, a conceptualization seems a promising approach. Fig. 2 shows the structural components of and relationships between enterprise supplier data. For reason of clarity, attributes were omitted in the figure. As the classifiers are enterprise specific, nine classifiers were selected. For each classifier a company specific value set was defined represented in a morphological field [17]. Tab. 2 shows the classifiers, the value sets, and instantiates them for the enterprise data element Identifier: VAT registration number. Selected values for the elements are colored in grey.

Conceptualizing Data in Multinational Enterprises: Model Design and Application

535

Fig. 2. Conceptual model for enterprise supplier data at EEE Table 2. Instantiation of classifiers for Identifier: VAT registration number View

Classifier Data definition Structural constraints Administration Data volume Change frequency Data ownership scope Governance Distribution type Storage Data provision Process diversity Usage Business processes

4

Value sets inter-divisional intra-divisional local none partial complete high medium low high low none inter-divisional intra-divisional local broadcast individual none central hybrid distributed high medium low support processes core & management processes

Summary and Outlook

The conceptual model for enterprise data describes structural and behavioral components of enterprise data elements and their relationships. Enterprise data classifiers represent business requirements to be met by enterprise data and need to be taken into account when designing enterprise data architecture. Interdependencies between enterprise data elements, between the views, or between individual enterprise data classifiers were not taken into account. Relations between enterprise data elements may involve concepts from entity relationship modeling and object oriented modeling like associations or inheritance. Concepts like multiplicity, direction, aggregation, and composition can be of relevance. Future research should analyze these interdependencies. The conceptual model for enterprise data is well-grounded in theory and practice and is applied to a single case study. In order to assess proper model design, an evaluation of the conceptual model against predefined research goals should be performed [18, 19]. The research contributes to the scientific body of knowledge by identifying, structuring and aggregating concepts of enterprise data architecture management. Future research on requirements analysis and strategic design of enterprise data

536

V. Ebner, B. Otto, and H. Österle

architecture can build on the conceptual model. In the practitioners’ community the conceptual model supports enterprise data architects in gathering and analyzing requirements to be met by enterprise data.

References 1. Loshin, D.: Master Data Management. Elsevier Science & Technology Books, Burlington (2009) 2. Dreibelbis, A., Hechler, E., Milman, I., Oberhofer, M., van Run, P., Wolfson, D.: Enterprise Master Data Management: An SOA Approach to Managing Core Information. Pearson Education, Boston (2008) 3. Otto, B., Schmidt, A.: Enterprise Master Data Architecture: Design Decisions and Options. In: 15th International Conference on Information Quality (ICIQ 2010), Little Rock (2010) 4. Dama: The DAMA Guide to the Data Management of Knowledge. In: Dama (ed.), 1st edn. Technics Publications, Bradley Beach (2009) 5. Periasamy, K.P.: The State and Status of Information Architecture: An Empirical Investigation, Orlando, FL (1993) 6. Yin, R.K.: Case Study Research. Design and Methods, 3rd edn. Applied Social Research Methods Series, vol. 5. Sage Publications, London (2002) 7. Schütte, R.: Grundsätze ordnungsmässiger Referenzmodellierung: Konstruktion konfigurations- und anpassungsorientierter Modelle. Gabler, Wiesbaden (1998) 8. Sowa, J.F., Zachman, J.A.: Extending and formalizing the framework for information systems architecture. IBM Systems Journal 31(3), 590–616 (1992) 9. Shankaranarayanan, G., Even, A.: Managing Metadata in Data Warehouses: Pitfalls and Possibilities. Communications of AIS 14, 247–274 (2004) 10. Levitin, A., Redman, T.: Quality Dimensions of a Conceptual View. Information Processing & Management 31(1), 81–88 (1995) 11. Bitterer, A., Newman, D.: Organizing for Data Quality. Gartner Research, Stamford (2007) 12. Khatri, V., Brown, C.V.: Designing Data Governance. Communications of the ACM 53(01), 148–152 (2010) 13. Leser, U., Naumann, F.: Informationsintegration - Architekturen und Methoden zur Integration verteilter und heterogener Datenquellen. Dpunkt. verlag, Heidelberg (2007) 14. Jung, R.: Architekturen zur Datenintegration. Gestaltungsempfehlungen auf der Basis fachkonzeptueller Anforderungen. Deutscher Universitäts-Verlag, Wiesbaden (2006) 15. English, L.P.: Improving Data Warehouse and Business Information Quality, 1st edn. John Wiley & Sons, Inc., New York (1999) 16. Porter, M.E.: Competitive Advantage: Creating and Sustaining Superior Performance. Free Press, New York (1998) 17. Ritchey, T.: Problem structuring using computer-aided morphological analysis. Journal of the Operational Research Society, 792–801 (2006) 18. Becker, J., Rosemann, M., Schütte, R.: Grundsätze ordnungsmäßiger Modellierung. Wirtschaftsinformatik 37(5), 435–445 (1995) 19. Gregor, S.: The Nature of Theory in Information Systems. MIS Quarterly 30(3), 611–642 (2006)

Suggest Documents