Page created: 4 Feb 2020 |
Page updated: 22 Jul 2020
Replication is not a general purpose synchronization facility as it creates replicas with exact copies of the replicated data. Synchronization, on the other hand, can transform data between two different Directory Information Tree (DIT) structures, map attribute types, synchronize subsets of branches and specific object classes. The differences between replication and synchronization are illustrated as follows:
- Replication cannot Synchronize between Different DIT Structures. The DN of replicated entries must be the same on all servers. In some situations, it may be desirable to replicate entries with the help of DN mapping that are under different base DNs, but represent the same data, for example uid=john.doe,ou=people,o=corp on one server may represent the same user as uid=john.doe,ou=people,dc=example,dc=com. This is not supported by replication. Synchronization fully supports this feature.
- Replication cannot Map Attribute Types or Transform Attribute Values. In some situations, it may be necessary to map attribute types or transform attribute values when synchronizing data from one server to another. Replication does not support either attribute type mappings or attribute value transformations.
- Replication does Not Support Fractional Replication. Replication cannot be configured to replicate a subset of the attribute types from the replicated data set. Synchronization fully supports this feature.
- Replication does Not Support Sparse Replication. Replication cannot be configured to replicate entries with a particular object class only. Synchronization fully supports this feature.
- Replication Requires Full Control of Replicated Data. When two servers participate in replication, both servers implicitly trust each other using public key cryptography and apply all updates received via replication, which is considered an internal operation. While trust between servers is established between two endpoint servers, synchronization does not require full control of the data. Disparate server system endpoints can be synchronized, such as a PingDirectory Server and a RDBMS database endpoint with each fully in control of its own data.