Data Integration - skrusche63/OASIS-ebXML-RegRep-v4.0 GitHub Wiki
The Association & Classification Information Model, both are component models of the OASIS ebXML Registry Information Model (ebRIM), are suitable tools to describe almost all application models or schemas in terms of the OASIS ebRIM.
The integration of heterogeneous data silos in a unified data environment is one of the application domains of the OASIS ebXML Registry Information Model v4.0. This model is a meta-model and is capable of describing application data as well as its related application models or schemas.
Traditional approaches to build a database system usually depend on 3 layers:
Application data (first layer) are described by application models or schemas (second layers), which are directly mapped onto the meta meta-model (third layer) of the chosen database management system.
A meta meta-model may be a
- Key Value Model
- Graph Model
- Relational Model or
- Object-oriented Model
In a data integration scenario we usually have to compare different application models or schemas e.g. for data cleansing. This is where meta-models such as the OASIS ebXML Registry Information Model (ebRIM) come into play.
From many data integration projects, where we used the OASIS ebRIM as an intermediate layer to describe different application models or schemas in parallel, we have learned, that is meta-model is capable almost any data description.
The subsequent image illustrates the usage of a meta-model from a conceptual point of view.
The image shows different independent database systems that are built on the basis of the 3 layer approach. In order to establish a unified data environment for these heterogeneous systems, a platform must be built from 4 layers with an additional meta-model layer.
From a more technical point of view such a unified data environment has 3 main building blocks:
- A commercial or open source database such as Cassandra (Key Value), OrientDB (Graph) or PostgreSQL (Relational).
- An OASIS ebIM profile for a chosen database. For graph databases, we have built profiles from version 4.0 of the information model for the NEO4J and OrientDB graph database. For the near future, it is planned to define a profile for the Cassandra NoSQL database.
- An application model or schema profile for the OASIS ebRIM. This a mapping description of how to describe the components of an application model or schema with the different component models of the OASIS ebRIM.
Our lessons learned from many data integration projects indicate that the Association Information Model in combination with the Classification Information Model is sufficient to describe almost all application models or schemas.
To be continued.