Node associations are represented by relationships. There are different types of relationships which define hierarchy, hosting, containment, ownership, manufacturer, category, versions, location, dependency, in fact, any kind of relationship required to model an IT infrastructure.
A type of relationship is represented as colon-separated strings. A relationship comprises more than the two node endpoints, it contains the "route" between the two. This is made up in the following way:
For example the relationship from a Host node to the Cluster node that contains the host is shown as:
The relationship from a Cluster node to a Host node that it contains is shown as:
A relationship is a bidirectional link between two nodes. In this section, the terms "start node" and "target node" are used to describe the way that the relationship is built. However, once the relationship has been built, the terms are redundant.
The general timeline for the creation of a relationship is:
- Create "start node".
- Create "target node".
- Create relationship between 1 and 2.
The individual components of the relationship are described in the following sections.
The node is any kind of node that can have a relationship, such as a Host node. All nodes can have relationships, including relationships.
The relationship kind describes the nature of the relationship. For example, the DeviceSubnet relationship kind describes the link between a device and the subnet that it is on, but without any directionality implied.
This section refers to Relationship Kinds as the colon-separated lists which describe how two nodes are related, the roles that each node plays in the relationship, and the Relationship Link which describes the nature of the Relationship.
A complete list of Relationship Kinds with their descriptions is provided in the table below.
Relates nodes into a hierarchy.
Relates a node to a file attachment.
Relates an element to a category it is in.
Relates an element to its owner.
Relates an employee to their manager.
Relates an element to the location it is in.
Relates a location to the larger location it is in.
Relates software to the host it is running on.
Relates hosts and software instances to contained virtual hosts.
Relates Hosts that appear to have changed identity
Relates an item of software to an item contained in it.
Relates an item of software to a service.
Relates software instances to communicating software instances.
Relates elements to elements they depend upon.
Relates an IP device to the subnet it is on.
Relates an IP device with one of its network interfaces.
Relates a network interface on a host to a port interface on a switch.
Relates a file to its host.
Relates a file to something that uses it.
Relates a World Wide Node Number (WWNN) to an HBA.
Relates a World Wide Port Number (WWPN) to a WWNN.
Relates the mounter and the FileSystem.
Directly Discovered Data
Relates a DiscoveryAccess to one of its results.
Relates a List of Members.
Relates sequential nodes.
Relates an inferred node with its source node.
Relationship to a status node.
Information about a discovery request.
Associates a pattern module to its contents.
Associates a pattern module to its pattern package.
Associates a pattern module to its dependents.
Pattern that is maintaining an inferred node.
Associates a pattern that will need the products of another.
Associate patterns that are being overridden or are overriding others.
Associated pattern has been deprecated or is deprecating others.
Errors generated by this pattern at runtime.
Associates a PatternExecution to its Pattern.
Associate a PatternTrigger with things it changed.
Associates a PatternExecution to its trigger nodes.
Associates a Pattern to the DDD nodes requested by it.
Relates the implementation of a resource
Relates the usage of a resource
Generic Modeling Relationships
Relates an element to its container.
Relates an element to details about it.
Relates Host and Software Instance nodes to SupportDetail nodes.
Relates an IntegrationProvider to all its created IntegrationPoints.
Containment of one node by another
Relates an IntegrationResult to the connection that was used to generate that result
Relates an IntegrationResult to the query that was used to generate that result
Associates a rule module to its dependents.
Associates a discovery access with an IP range for revisiting purposes.
Foundation User Relationships
A Favorite item for a User
Relates to the invoking user
Hardware Reference Data Relationships
Relates physical hosts to hardware reference data
The role describes the part that its node plays in the relationship.
For example, where a Host node represents a host which is running a software process:
- A Software Instance has been created which represents the process running on the host.
- The Host node is acting in the role of Host; the host for some running software.
- The Software Instance is acting in the role of RunningSoftware; software which is running on a host.
Roles are required because a node can play one of many parts in a relationship, and clarification is needed; the role clarifies the part that the nodes are playing in the relationship.
A complete alphabetical list of Role Kinds in the default taxonomy with their descriptions is provided in the table below.
Node containing an attachment.
Business owner of an element.
Category of elements.
Child in hierarchy.
An element contained within another.
Host contained within another host.
A location inside another location.
Piece of software contained inside other software.
An element containing others.
Entity that depends upon another.
Entity depended upon by another.
A detail belonging to an element.
IP device belonging to a subnet.
A device with an interface, for example, a network interface.
A device with an IP address.
Element belonging to a category.
Element in a location.
Element using a file.
Element with associated details.
Element containing a collection of elements.
Element with an associated status.
Element being managed.
A range used to control endpoint access.
A Fibre Channel HBA with a WWNN.
A Fibre Channel WWNN for a HBA.
A Fibre Channel WWNN with a WWPN.
A Fibre Channel WWPN.
Physical host to which hardware reference data will be related.
Support Detail Data is for Hardware.
Home location of an element.
Host for Software Instances.
Host containing other hosts.
A File on a Host.
IT owner of an element.
Interface of an IP device.
Interface with an IP address.
An IPv4 address.
An IPv6 address.
Current location of an element.
A location containing other locations.
Manufacturer of an item.
Member of a list or collection.
Mounted file system.
Mounter of a file system.
One end of a network connection.
A new node.
A replaced node.
Next node in sequence.
Operating system of a host.
Support Detail Data is for Operating System.
Item owned by someone or something.
Owner of an item.
Parent in hierarchy.
Pattern definitions function.
Previous node in sequence.
Holds hardware reference data.
Resource being used.
User of resource.
Software running on a host.
Aggregate software running on a host.
Service being provided.
Provider of a service.
Piece of software containing other software.
Support Detail Data is for Software.
Element containing storage.
Storage contained in an element.
Storage for an element.
Subnet of an IP device.
Person or group responsible for supporting an element.
A pattern that is overridden by another.
A pattern that is overriding another.
An ECAError related to a pattern.
A pattern with ECAErrors.
By linking data items and their attributes to the evidence for them, BMC Atrium Discovery enables its data to be easily verified; a prerequisite for trusting it. This feature is called Provenance.
Provenance Data is created by the Reasoning Engine and consists of relationships between Directly Discovered Data items and entities inferred from them - a relationship back to the source from the inferred entity. For example, there is a provenance relationship between each Software Instance (SI) and the data that caused the SI to be inferred - typically one or more processes, files, or the like.
Provenance information is meta-information describing how the other information came to exist. It is generated as Reasoning builds and maintains the model. Provenance information is stored on relationships in the model.
- Inference Types
- Primary — indicate that the existence of the evidence node is the reason that the inferred node was created.
- Contributor — indicate that the evidence node provided information used in building the inferred node.
- Associate — indicate that BMC Atrium Discovery knows there is a relationship between the evidence node and the inferred node.
- Relationship — indicate that BMC Atrium Discovery knows that a relationship exists because of the evidence node.
- Deletion — indicate that the removal of the inferred node was due to withdrawal of the evidence node.
- Maintaining Pattern — the pattern maintaining a node.
If you need to access the data in your model you should access the UI functions provided in BMC Atrium Discovery that enable this. Apart from specialist purposes, you should not need to understand the provenance implementation details in BMC Atrium Discovery. However, an overview is provided below.
Provenance Implementation Details
To represent provenance information, attributes can be stored on the Inference relationships. The keys attribute contains a list of the attribute keys that were set or confirmed on the inferred node as a result of the inference. The rule attribute contains the name of the reasoning rule that decided upon the inference. Provenance information is only created and is not updated.
Inferred nodes are related to DDD nodes using 'Creation Inference' relationships. The Inferred node has the role InferredElement; the DDD node has the role Primary, Contributor or Associate as appropriate. To support the full lifecycle of inferred nodes there are complimentary Removal Inference relationships. Destruction is the complimentary role to Primary, and Withdrawal is the complimentary role to Contributor.
The method of linking of discovered information to inferred information means that in general, each inferred node will have relationships to many discovery data nodes, representing multiple discoveries of the entity over time.
Relationships between Relationships
Relationships are technically a sub-class of nodes and they have the same basic characteristics. Relationships can have attributes stored in them and can have relationships to other nodes in the same way that nodes can.
A relationship connects two nodes between their role, however a relationship can also act in a role in a relationship. Therefore, you can have relationships between nodes and relationships or relationships between relationships.
Relationships between Inferred nodes
Using patterns, the reasoning engine can construct a number of relationships between inferred nodes, as follows:
Links an SI to the Host it is running on.
A client-server relationship between SIs. This can be inferred based on the presence of network connections between processes. It can also be based on other evidence such as configuration files or directly querying the endpoints.
Equivalent to the client-server case, but where the endpoints in a communication are peers, rather than client and server.
A dependency between SIs, other than a communication link. Attributes on the relationship can indicate the kind of dependency.
Represents SIs that logically contain other SIs.
Links a virtual host to the Software Instance that implements it.
Links a configuration file to a Software Instance.
Represents the SIs that make up a BAI.
Represents BAIs that logically contain other BAIs.
A dependency between BAIs.
Indicates the Hosts which form a cluster.
Indicates the SIs that provide clustering services for a cluster.
Represents the relationship between logical hosts and their container for Sun E15Ks and similar.
Represents the relationship between network interfaces and the subnets they are connected to.
In addition to these relationships, the reasoning engine also constructs relationships to the Pattern Management nodes, for example, patterns.