Abstract
Data integration is one of the core responsibilities of EDM (enterprise data management) and interoperability. It is essential for almost every digitalization project, e.g., during the migration from a legacy ERP (enterprise resource planning) software to a new system. One challenge is the incompatibility of data models, i.e., different software systems use specific or proprietary terminology, data structures, data formats, and semantics. Data need to be interchanged between software systems, and often complex data conversions or transformations are necessary. This paper presents an approach that allows software engineers or data experts to use models and patterns in order to specify data integration: it is based on data models such as ER (entity-relationship) diagrams or UML (unified modeling language) class models that are well-accepted and widely used in practice. Predefined data integration patterns are combined (applied) on the model level leading to formal, precise, and concise definitions of data transformations and conversions. Data integration definitions can then be executed (via code generation) so that a manual implementation is not necessary. The advantages are that existing data models can be reused, standardized data integration patterns lead to fast results, and data integration specifications are executable and can be easily maintained and extended. An example transformation of elements of a relational data model to object-oriented data structures shows the approach in practice. Its focus is on data mappings and relationships.
Highlights
The advantages are that existing data models can be reused, standardized data integration patterns lead to fast results, and data integration specifications are executable and can be maintained and extended
The application of data integration patterns (DIP) in conjunction with enterprise integration pattern (EIP) shows that it is possible to provide a framework for complex data integration services that reduces the manual programming effort
This proof of concept implementation of pattern and model-based data integration is promising. The advantages of such an approach are: (1) a common language for system and data integration is introduced, that leads to better communication between the stakeholders of an integration project, (2) the data integration tasks are defined formally in a precise and concise manner with a visual notation (UML) eliminating the need for further documentations, and no programming knowledge is necessary because the data engineer is able to define and execute powerful data integration tasks without manual coding through code generation
Summary
Enterprise software systems such as ERP, CRM (customer relationship management) and eCommerce, in addition to IoT (Internet of Things) and EDI (electronic data interchange) applications deal with simple or complex data, either in the form of operational data in the context of OLTP (online transaction processing) or analytical data during OLAP (online analytical processing) [1]. A data warehouse for analytics or a relational database management system (RDBMS) that serves as an operational persistence layer and data hub for external eCommerce systems are examples of software applications that need data integration. This, and the fact that in probably every digitalization project in practice, data exchange and integration is an important topic that has led to the development of a data integration pattern (DIP) catalog and the model-driven and pattern-oriented approach for data integration that is described here.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
