Persistent Roles

When a persistent role is used, its instances are written to a separate table. You can use persistent roles to avoid performance problems when generic accesses to relationships are made.

If the relationship type is not entered when the relationships are read, performance problems can occur because the relationship records are distributed to multiple tables. If an object can occur in several relationships and several roles, it is possible that the system must search for entries in a large number of relationship tables in all of these roles, even if the object does not occur in any relationship at all.

You can avoid this problem by using persistent roles. It is then adequate to find an access to all roles in which the object occurs. The relationship record can be searched in a targeted way in the relationship tables which use the corresponding role.

As a result of the persistent roles, it is possible that there will be a very large table with redundant information. You therefore have to avoid persistent roles if relationships (in which a role type occurs) are not to be accessed generically, but rather only with the entry of the relationship type.

The following table shows an overview of the accesses when reading related objects with and without persistent roles. Here, the ideal situation that no accesses to tables of the previous service are necessary is used. For comparison purposes, the accesses for the relationship service used before Release 6.10 are additionally given.