...
Jira | ||||||
---|---|---|---|---|---|---|
|
...
Forward to Area 1 - Assets and Connectors ==>
In the Atlas build, model files are added in alphanumeric order to build up the type library. Therefore the model files are numbered to ensure its dependencies have been resolved before a model file is loaded. For the open metadata types, the model file number is shown in package name to document the intended load order. The numbering scheme of the metadata areas described in Building out the Apache Atlas Typesystem provides the high-level numbering system. For example, area 1 has models 0100-0199, area 2 has models 0200-299, etc. Each area's sub-models are dispersed along its range, ensuring there is space to insert additional models in the future.
In addition to the 7 metadata areas, the open metadata models introduces some base definitions and structures that used throughout the open metadata models. These need to be loaded first and so we have added an Area 0 that includes these base definitions. The current models defined for Apache Atlas have been moved to 1000. Thus the models numbers from 0000 to 0099 will look something like figure 1.
Figure 1: Summary of Area 0's packages |
Details of the new models are shown below. The original Apache models are linked to on page: Atlas Model.
Existing Base Model
The existing base model appears at this point defining key concepts such as Referenceable, Asset, Infrastructure, Process and DataSet.
Figure 2: Existing base model |
Linked Media Types
Linked media types describes the simple structures that are used repeatedly in open metadata to connect it to documents and entities in other types of repositories.
Figure 3: Linked media for any referenceable entity |
External Identifiers
External Identifiers are identifiers for a Referenceable that are used in external systems.
Figure 4: Correlating metadata with information in other systems |
Property Facets
Property facets allow any entity to be extended with additional properties. This is particularly useful for storing metadata that originated in another type of metadata repository, or tool since it allows vendor/tool specific values to be stored.
Figure 5: Adding custom properties to any referenceable entity |
Locations
It is important to understand where assets are located to ensure they are properly protected and comply with data sovereignty laws. The open metadata model allows location information to be captured at many levels of granularity.
Figure 6: Understanding where data assets and services are located | The NestedLocation relationship allows hierarchical grouping of locations to be represented. Notice that locations can be organized into multiple hierarchies. The AdjacentLocation relationship links locations that touch one another. The notion of a location is variable and the classifications PhysicalLocation, SecureLocation and MobileLocation help to clarify the nature of the location.
|
Hosts and Platforms
The host and platform metadata entities provide a simple model for the system infrastructure (nodes, computers, etc) that data resources are hosted on.
Figure 7: Defining the platform that the data assets and services run on | The host can be linked to its location through the HostLocation relationship. |
Complex Hosts
The complex hosts handle environments where many nodes are acting together as a cluster, and where virtualized containers (such as Docker) are being used.
Figure 8: Supporting server clusters and server virtualization (server containers) | A HostCluster describes a collection of hosts that together are providing a service. Clusters are often used to provide horizontal scaling of services. A VirtualContainer provides the services of a host to the servers deployed on it (see 0040 below). When the server makes requests for storage, network access etc, the VirtualContainer delegates the requests to the equivalent services of the actual host it is deployed on. VirtualContainers can be hosted on other VirtualContainers, but to actually run they need to ultimately be deployed on to a real physical Host.
|
Servers
Servers describe the middleware servers (such as application servers, data movement engines and database servers) that run on the Hosts. Within the server model we capture its userId. Most metadata repositories are run in a secure mode requiring incoming requests to include the requester’s security credentials. Therefore we have an identifier for each unique logged on security identity (aka userId). This identity is recorded within specific entities and relationships when they are created or updated. By storing the user identifier for the server, it is possible to correlate the server with the changes to the metadata (and related data assets) that it makes.
See model 0310 Actors in Area 3 for details of how user identifiers are correlated with people and teams).
Figure 9: Servers and their connectivity and capabilities | Open metadata may also capture the network endpoint(s) that the server is connected to and the host it is deployed to. The endpoint defines the parameters needed to connect to the server. It also features in the Connection model used by applications and tools to call the server. Thus through the endpoint entity it is possible to link the connection to the underlying server. Within the server are many capabilities. These range from full applications (see 0060 below) to security plugins to logging and encryption libraries. Different organizations and tools can choose the granularity in which the capabilities are captured in order to provide appropriate context to data assets and the decisions made around them. |
Networks and Gateways
The network model for open metadata is very simple, to allow hosts to be grouped into the networks they are connected to. This can show details such as where hosts are isolated in private networks, where the gateways onto the Internet.
Figure 10: The networks that specific hosts connect to |
Cloud Platforms and Services
The cloud platforms and services model show that cloud computing is not so different from what we have been doing before. Cloud infrastructure and services are classified as such to show that the organization is not completely in control of the technology supporting their data and processes.
Figure 11: Cloud platforms and services | The cloud provider is the organization that provides and runs the infrastructure for a cloud service. Typically the host it offers is actually a HostCluster. The cloud provider may offer infrastructure as a service (IaaS), in which case, an organization can deploy VirtualContainers onto the cloud provider's HostCluster (see model 0035 above). If the cloud provider is offering platform as a service (PaaS), an application may deploy server capability onto the cloud platform. If the cloud provider is offering Software as a Service (SaaS) then it has provided APIs backed by pre-deployed server capability that an organization can call as a cloud service.
|
to ==> Area 1 - Collaboration
...
...