Abstract
Even though relational database systems still play a major role in data management, they are reaching their limits in some cases. This relates, for example, to the increasingly large, often unstructured or highly linked data that need to be managed. To meet these requirements, various specialized NoSQL database systems have been developed. Their specialization has them differ significantly in ...
Abstract
Even though relational database systems still play a major role in data management, they are reaching their limits in some cases. This relates, for example, to the increasingly large, often unstructured or highly linked data that need to be managed. To meet these requirements, various specialized NoSQL database systems have been developed. Their specialization has them differ significantly in terms of data model, query language and design guidelines. In addition, applications sometimes have complex requirements that cannot be met by a single database system ("one size does not fit all"). As a result, polyglot persistence and multi-model database systems become increasingly important. Due to the large number of specialized database systems and changing requirements for the database system used by an application, there is also a need to be able to easily migrate from one database system to another. Here, the heterogeneity and schema flexibility (most NoSQL systems are "schema on write") pose some challenges in this regard. To avoid point-to-point transformations, a platform-independent metamodel that combines the various platform-specific models plays a central role. This also includes the necessary mapping rules between the metamodel and the platform-specific models. In this paper, different metamodels are evaluated in accordance with previously defined criteria and a suitable metamodel (the M-Model) with a focus on migration between different database systems is developed.
Altmetric