Automatic grouping

When you baseline a data center, one of the first tasks is to subdivide it into small groups of hosts. BMC Discovery does this automatically for all discovered hosts at the end of each scan.

Automatic grouping is intended as a baselining tool for use at the scale of 100 to 500 hosts. At larger scale, the usefulness and usability of the visualizations is diminished.

How automatic grouping works

Three types of group are created automatically:

• Group around a host
• Clients of a host
• Hosts with no relevant communication

When a host is involved in automatic grouping, the Host:Host:ObservedCommunication:Host:Host relationship (Communicating With) is displayed on the host page. 

These groups are created in the following way after each scan:

• Groups hosts which communicate with one another—These are called group around hostname.
  Any two hosts that are seen to communicate are called neighbors. When a pair of hosts is considered, the connection between them is ranked. A higher rank increases the probability of them being grouped. The ranking is calculated according to the proportion of the total neighbors that they share. Where more neighbors are shared, the ranking is higher, and the hosts are more likely to be grouped. This is used to reduce the chances of clients of shared infrastructure being grouped on the basis of those connections.
• Groups hosts that are clients of a particular set of servers—These are called clients of host.
• Hosts with no observed communication—These are grouped in the Hosts with no relevant communication group. Additional hosts that communicate only with hosts that have been excluded from grouping are also placed in this group.

Host groups are stored as AutomaticGroup nodes in the datastore.

To see how your hosts are grouped

There are several ways to view automatic grouping.

Automatic grouping channel

The automatic grouping channel shows a summary of the current grouping results. Click any of the icons to view a visualization. The top row shows:

• The overview—This is a visualization that shows all automatic groups and how they communicate with each other.
• No relevant communication—A visualization of the group containing hosts that have either no communication at all or are communicating only with excluded hosts.
• Excluded hosts—Hosts that have been excluded from the automatic grouping process. 

The next row shows up to five of the most highly connected groups; that is, those that are communicating with the most other groups. The final row shows up to five of the largest unconnected groups, self-contained groups that do not communicate outside themselves at all.

Related topics

• Auditing a data center