This topic describes the COSMOS Reconciliation Taxonomy.
Is your organization's IT management data scattered in multiple repositories, even when the data are about the same resource? For example, you may have data about the same computer in a configuration management database (or two), an IT asset tracking database, perhaps in a service desk for incident and problem management, as well as that physical inventory database Joe created before he retired. The data would be more useful if the data are integrated, but integration will be complicated by the fact that each repository probably uses a data model different from the others. It would not be surprising if you don't even know which repositories have data about the same resource.
One way to approach this integration challenge is in two steps.
The Reconciliation Taxonomy (RTx) helps with the first of these steps. The RTx, sometimes known as a federation reconciliation catalog, is published by the Eclipse COSMOS project. It defines several resource types, known as facets in RTx, and widely used identification properties for each facet, that have a good probability of enabling automatic reconciliation of resource identities, and providing sufficient information for management tools and IT personnel to correctly locate/recognize each resource.
The Reconciliation Taxonomy defines a basic set of resource types and the properties typically used to identify instances of each resource type. The purpose of the taxonomy is to describe enough information about a resource (e.g., a computer system) to determine whether two management data repositories have data about the same resource, even given that each repository uses a different data model. For example, some identifying properties about a computer system are manufacturer, model, serial number, system board UUID, and primary MAC address. Such a taxonomy could be used in different contexts, including implementations of the CMDB Federation specification and launch in context scenarios.
The use of a reconciliation taxonomy is a compromise between trying to get the entire IT industry to adopt a single data model (considered to be impractical) and defining no common data at all, which would require a bespoke model mapping for each pairwise integration of management data repositories.
Management tools using the CMDB Federation (CMDBf) specification are one example of potential uses. The CMDBf specification was originally published by a six-company consortium; the Distributed Management Task Force (DMTF) is carrying the work forward and has published two works in progress, the latest of which is accessible here.
The accompanying picture is an example of identity reconciliation using the RTx. Two management data repositories, MDR#1 and MDR#2, each have data about a computer, and each know a subset of the identifying properties contained in the RTx for computers. When each registers with a Federating CMDB the list of computers it knows, the CMDB inspects the identifying properties of each computer to determine if it already knows about this computer. Because both MDRs know the primary MAC address, the CMDB recognizes that each MDR has data about the same computer, so it creates only one object for the computer. Typically it would maintain links back to the data in each MDR. The CMDB could respond to queries by returning links to the data in each MDR and/or marshalling data from each MDR together (e.g., if it has done Step 2 and knows how to map at least some of the non-identifying properties) and returning a composite view.
Another potential use of the RTx is operational integration between management tools that use different data models. If each understands the RTx, one could ask the other to perform an operation on a resource by passing parameters for the RTx facet and its identifying properties. The receiver would analyze the parameters and, if it could resolve them to a known resource, perform the operation on that resource. A related use case is launch in context between management tools.
The RTx is described in one XML Schema Definition (XSD) file and one informative document. The informative document further describes the information in the XSD and provides examples of its use.
To load the RTx into Eclipse, navigate to HEAD/org.eclipse.cosmos/resource-modeling/org.eclipse.cosmos.rm.example.rtx in the Eclipse CVS repository.
The RTx intentionally started with only twenty seven resource facets and eighteen relationship facets. A motivation for publishing and developing it in COSMOS, at least initially, is to use a pragmatic experiential approach to select which types and identifying properties to include. It is expected that over time more types and more identifying properties for each type will be added as additional use cases are addressed. Looking down the road, a standards body may start to work in this area, possibly using the RTx as input.