The Open Metadata Repository Services (OMRS) enable metadata repositories to exchange metadata irrespective of the technology, or technology supplier.
Traditional metadata management technology tends to centralize metadata into a single repository. An organization often begins with a single metadata repository, typically deployed to support a single project or initiative. However, over time, depending on the tools they buy, the projects they run or the political structures within the organization, the number of deployed metadata repositories grows, creating multiple metadata silos. So for example, an organization may have:
-
A metadata repository and tools for its governance team. This metadata repository may host the canonical glossary, and the governance policies, rules and classifications.
-
A metadata repository for its data lake. This metadata repository has the details of the data repositories in the data lake and the movement of data between them.
-
A metadata repository for its data integration tools that continuously extract data from the operational systems and sends them to the data lake.
The role of the OMRS is to bring these metadata repositories together so this metadata can be linked and used together across the organization. It enables these metadata repositories to act as an aggregated source of metadata. The metadata repositories using OMRS may be a mixture of repositories from different vendors that support the OMRS integration interfaces.
The OMRS supports peer-to-peer operation. This means there is an instance of the OMRS running with each metadata repository. This OMRS instance acts as the metadata repository's interface to the wider open metadata ecosystem. This includes distributing metadata to other repositories through the event bus and supporting metadata API requests.
An instance of the OMRS is communicating with other OMRS instances, each located in different metadata repositories. The collection of metadata repositories communicating via their local OMRS instances is called an Open Metadata Repository Cohort.
One of the principles of open metadata is that metadata should be managed as close to its source as possible but it should also be easily accessible through standard open APIs and notifications.
Another principle is that only one repository has write access to a specific piece of metadata. Other copies of this metadata are read-only.
Taking these two principles together, an instance of the OMRS aims to store any new metadata it receives in its local repository. Only if the local repository is not able to store it, will the OMRS seek an alternative location for it. The local OMRS then supplements this metadata with read-only copies of metadata from other repositories that are of interest to its local users.
Whichever repository is first used to store a piece of metadata, becomes its home repository. The home repository is the only repository able to update this metadata. The read-only copies stored in other repositories are called reference copies.
Egeria's meta-model defines the standard way for metadata to be represented and communicated, fundamentally consisting of Entities, Classifications and Relationships.
See the documentation.
The code for OMRS is organized into five modules:
-
repository-services-apis contains the Java client interfaces, connector interfaces and event structures for the repository services.
-
repository-services-archive-utilities contains the common utilities used to build and process Open Metadata Archives.
-
repository-services-client contains the Java interfaces and for calling remote repository services.
-
repository-services-implementation contains the support for the peer-to-peer metadata exchange and federation.
-
repository-services-spring uses spring to create the OMRS REST services.
Return to open-metadata-implementation.
License: CC BY 4.0, Copyright Contributors to the ODPi Egeria project.