BMC Atrium Discovery is based on a datastore that is designed to handle very large, complex and interconnected data models. The datastore defines data as a set of objects and the relationships between them.

Using objects and attributes

An object is any data entity of a specific type, or kind, that has characteristics defined by attributes. The types of object permitted and their attributes are defined in the taxonomy. For example, an object of kind Host has attributes including Processor, Domain, and Operating System. Some attributes have a predefined list of valid values, or are of a specific type, such as a date, a Boolean value, or a list (such as a number of associated IP addresses). However, many attributes are defined as strings with no size limit, allowing very flexible data definitions.

Attributes can have one or more of the following flags:

• Required: Specifies that an attribute is mandatory when data is created. Required attributes are marked by an asterisk on the user interface.
• Expected or Optional: Defines whether or not an attribute affects the Data Quality of an object. Note that unless the attribute is also marked Required, an Expected attribute is not mandatory.

How the datastore uses history data

History information about each object is recorded in the datastore. This audit trail includes details of the changes made, the date and time of the changes, and the user who made the changes. Destruction of the data is also audited, where destroyed data is marked as virtually destroyed rather than completely removed. You can view the history of each object in the datastore, and you can search for and display details of destroyed objects. For details about how you can view historical comparisons in BMC Atrium Discovery, see Comparing the history of nodes.

For information on how history is purged, see Purging the history.

Associating file attachments

You can associate files of any type with a managed element to provide additional or supporting information.

A managed element is any one of the following:

• Host
• Software Instance
• Business Application Instance

These files are uploaded to BMC Atrium Discovery for easy access in the event of a disaster recovery.

Any number of attachments can be associated with a single object, and the same file can be attached to multiple objects. A copy of the attached file is uploaded to the BMC Atrium Discovery appliance for secure storage. You can subsequently view a file or update it at any time. There are no restrictions to the type of file, because BMC Atrium Discovery stores the attachments and does not process them. Attachments are assigned to a particular category, such as Audit or System Recovery. The BMC Atrium Discovery Administrator can set up additional categories. For more information, see Setting up standard data categories.

Maintaining the quality of data

Because the data that is stored by BMC Atrium Discovery might be incomplete at a given time, an indication of data quality is maintained and reported for each object. The data quality report includes a method for weighting data problems so that an issue defined as serious will have a greater effect on the overall data quality for that object than a minor issue.

To calculate the data quality for an object, each potential data problem (such as a missing attribute value or relationship) is identified and quantified. Each possible problem is assigned a numerical value between 0.1 and 1, where complete and accurate data has a value of 1. The values are used to provide an overall data quality value, which is assigned to one of three categories:

• Good: Data Quality Indicator calculated as over a given high threshold.
• Medium: Data Quality Indicator calculated as between the high and low thresholds.
• Poor: Data Quality Indicator calculated as below the low threshold.
  A background color to an icon indicates the data quality of that object on the View Object and Report pages.

Acquiring data

The data used by BMC Atrium Discovery can be acquired from the following data sources:

• Discovery: Discovery of your infrastructure is the best way of acquiring information about the hardware in your system and the programs running. For details, see The Discovery Engine.
• The Pattern Language: Data is created using the Pattern Language. The patterns trigger on certain conditions and can create data which models different systems/configurations, for example, Solaris Zones and Virtual Machines. For more information on managing patterns, see Pattern management.
• Import: Data about your current systems and business can be imported from existing systems such as Asset Management or Inventory systems and from databases.