Identification Service

The Identification service uses identification rules to identify CIs. An identification rule is composed of three criterion sets:

  • Identification criteria. A set of criteria that defines all possible conditions to find all candidate CIs for matching to a newly introduced CI. Two CIs are identical if one of the identification criteria is satisfied.

  • Match criteria. There are two types of match criteria:

    • Match verification criteria. A set of criteria that are applied to all candidates left over after performing identification (in the previous step). Match verification ends successfully only when all applied verification criteria are true or missing data (NA).

    • Match validation criteria. An ordered set of criteria that is applied to all candidates left over after performing match verification. For each criterion, the following results are possible:

      • A true result implies a match

      • A false result implies no match

      • Missing data (NA) causes reconciliation to proceed to the next criterion. If all validation criterion are NA, then all candidates left after match verification will be implied as matched

When trying to determine whether or not two CIs are the same (according to the identification rule), each criterion is evaluated. Each criterion, in one of the above sections, contains one or more conditions. There are two types of conditions:

  • Attribute conditions. Test whether or not two CIs have the same value (or same value while ignoring case) as the given attribute defined for the condition.

  • Connected CI conditions. Test whether or not the two CIs share the same number of connected CIs defined for the condition.

For a criterion to evaluate to a true value, all of the conditions must evaluate to true. If one condition evaluates to false, the entire criterion evaluates to false. If one condition cannot be evaluated (its required data is missing in one of the two CIs being identified), the condition returns unknown, and, in turn, the entire criterion that contains it returns as unknown.

The identification process is as follows:

  1. At least one identification criteria must return true. The identification process tests the identification criteria.

    • If there is at least one identification criterion that returns true, the identification process continues.

    • If no identification criterion returns true, then the CIs are not the same and the identification process ends here.

  2. Are there contradictions between the CIs? The identification process tests the verification criteria.

    • If all of the verification data returns either true or unknown, the identification process continues.

    • Otherwise, there is a contradiction between the two CIs, and they are not the same and the identification process ends here.

  3. Evaluate the validation criteria. The identification process tests the verification criteria one by one and according to each one's priority:

    • If the validation criterion is satisfied, the two CIs as the same.

    • If none of the validation criteria are satisfied, and a validation criterion returns false, the two CIs are unmatched.

    • If none of the validation criteria are satisfied or refuted:

      1. The the criteria from the next priority are evaluated.
      2. If there are no more criteria (the last priority was evaluated), then the two CIs are the same.

Identification Process Example

The following items are used in this example:

Input node
  • name = n1
  • ip_address = ip1
  • MAC address = m1
  • os_family = nt
UCMDB nodes
  • N1: name=n2

  • N2 :

    • ip_address=ip1,ip2
    • MAC address=m1
    • os_family=nt
  • N3:

    • name=n3
    • MAC address = m1
    • hw_id = id1
    • os_family = unix
  1. For each UCMDB node, run the identification criteria:

    • Node N1 does not match any identification criteria, so it will not be added to the candidates list.

    • Node N2 matches the IP identification criterion of the input node, so it will be added to the candidates list.

    • Node N3 does not match the input node by the IP identification criterion, but does match by the MAC address identification criterion, so it will be added to the candidates list.

    Therefore, the candidates list is: N2 and N3.

  2. For each node in the candidates list, run OS match verification criteria. Node N3 does not satisfy this rule, since its OS is UNIX and the input node’s OS is NT. Therefore, N3 will be removed from the candidates list.

    The candidates list is: N2.

  3. Run the match validation criteria one by one on node N2. Since node N2 has no data conflicts, the match validation criteria are approved and N2 is marked as matched.

    The result of the identification process is: N2 = ip_address=ip1,ip2, MAC address=m1, os_family=nt.

To configure match criteria, see Identification and Match Criteria Configuration .

For examples of configuring identification rules, see Examples of Identification Configuration.

Note The identification process creates an identification map which is limited by default to 2,000,000 mappings. Each identification between two CIs is considered a separate mapping for each CI involved. Additionally, there is default limit of 5,000 mappings per CI. If you get an error saying that the identification map is too large, it may indicate that there is a problem with your data (for example, many nodes connected to a single IP address). Check the cmdb.reconciliation.identification.log for warnings. If there are none, check the cmdb.reconciliation.log to examine the actual identification map for unusual items.

Related Topics Link IconRelated Information