Identity Manager 8.1 - Configuration Guide

Process tracking for DBQueue Processor operations

In order to track inherited calculations as a result of changes to the system, the GenProcID is always passed to the DBQueue Processor operation. There may only be one entry in the DBQueue for each operation and object in case of follow-on operations. To map such processes, a new GenProcID is issued and used in subsequent processes. The conflicting processes and their GenProcID’s are saved in the table DialogProcessSubstitute.

When a new GenProcID is created for conflicting processes, the following rules apply:

• Several of the same DBQueue Processor operations on one object are merged into one process (one GenProcID). This uses existing substitute processes if the number is identical to the predecessor (with respect to the root processes).
• If further conflicts occur in the sequence, the GenProcIDs that have already been replaced are reset to the original and a new substitute is created.
• A substitute is only valid for one set of original processes.

Related Topics

• Example of replacing the GenProcID
• Processing DBQueue tasks

Example of replacing the GenProcID

A hierarchical role structure exists which consists of 4 roles O1, O2, O3 and O4. Employee X is assigned to roles O1, O4 and O3. The assignment of applications to roles is depicted in the following:

Figure 33: Role structure as in the example above

Three processes run between two DBQueue Processor executions, each with its own GenProcID:

• P1: Application A1 is assigned to the role O1
• P2: Application A2 is assigned to the role O1
• P3: Application A3 is assigned to the role O2

The following operations are in the DBQueue (table DialogDBQueue) and in the process information:

The operation OrgHasApp cannot be subdivided with respect to O1 because the union quantity of the applications is not yet computed. At this point, no more information is available as to which GenProcID has been entered by the assignment for which application.

In order to achieve uniqueness for the combination of operation and object, a new GenProcID P4 is introduced and the two O1 operations are compacted into this GenProcID. P1 and P2 are noted in the table DialogProcessSubstitute as possible predecessors of P4 (but not clearly in the individual actions).

The following constellations can occur depending on whether the operation OrgHasApp is processed as a single step or in bulk:

Case 1) O1 is calculated and then O2.

Case 2) O2 is calculated and then O1.

Case 3) O1 and O2 are calculated together simultaneously in a bulk operation.

After these operations have been executed and assuming that they all cause changes to the total sets affected, the following situation arises:

Case 1) O1 is calculated and then O2.

Before the next DBQueue Processor run, the GenProcID’s must be compressed again, because the OrgHasApp operation did not produce a unique result for the object O2. P5 is introduced with possible predecessors P4 and P3.

Now the calculation is done for O2:

Because O3 is not unique, P6 is introduced with possible predecessors P4 and P5.

After O3 and O4 have been calculated, the following situation exists:

There is no uniqueness for object X such that P7 is introduced with possible predecessors P4, P5 and P6.

Case 2) O2 is calculated and then O1.

After execution the following entries are in the DBQueue:

The following situation is the result after the next step:

To achieve uniqueness for O3 a process P5 with possible predecessors P3 and P4 is created:

After the calculations, the following situation exists:

There is no uniqueness for object X such that P6 is introduced with possible predecessors P4 and P5.

Case 3) O1 and O2 are calculated together simultaneously in a bulk operation.

After the first step in the calculation the following entries are in the DBQueue:

Uniqueness is achieved for O3 by introducing P5 with possible predecessors P3 and P4:

After the next step in the calculation, the following content is found

After O3 has been calculated in the next run and has not created a new PersonHasApp entry, only X exists with P4 and P5 because X already exists with P4.

There is no uniqueness for object X such that P6 is introduced with possible predecessors P4 and P5.

Archiving and deleting records

All records logged in One Identity Manager are initially saved in the One Identity Manager database. The proportion of historical data to total volume of a One Identity Manager database should not exceed 25 percent. Otherwise performance problems may arise. Records should be regularly removed from the One Identity Manager database and archived.

The following methods are provided for regularly removing recorded data from the One Identity Manager database:

• The data can be transferred directly from the One Identity Manager database into a One Identity Manager History Database. This is the default procedure for data archiving. Select this method if the servers on which the One Identity Manager database and the One Identity Manager History Database are located have network connectivity.
• The data is deleted from the One Identity Manager database after a certain amount of time without being archived.

For detailed information about setting up archiving of data in a History Database, see One Identity Manager Data Archiving Administration Guide.

Detailed information about this topic

• Direct deletion of records in the One Identity Manager database
• Specifying data retention periods

Direct deletion of records in the One Identity Manager database

If records from a subsection should be kept in the One Identity Manager database for a certain amount of time but are not archived later, you then have the following options:

• To exclude subsection from archiving do not configure it for export, but only specify a retention period.
• To delete all subsections with archiving, specify the retention period. Enable the Common | ProcessState | ExportPolicy configuration parameter in Designer and enter the value NONE.

The records are deleted from the One Identity Manager database by the DBQueue Processor when the retention period has ended. In addition, all entries for triggered actions that have no corresponding records in the subsections are deleted.

If there is a large amount of data, you can specify the number of objects to delete per DBQueue Processor operation and run in order to improve performance. You use configuration parameters to make the choice for each subsection.

Related Topics

• Specifying data retention periods