Progress
Database Administration
Guide and Reference


Database Ownership Models

Data ownership models determine how changes to a database (or site) affect other databases in the network. This section describes three models, data distribution, data consolidation, and the peer-to-peer model, and how they relate to replication.

Distribution Model

In the distribution ownership model, a single master database owns the data. The master database is the read/write area, and all changes are made to this database only. All changes are then propagated to the remote sites in a read-only state. The remote sites cannot change the data, only view it. In terms of replication, the chief advantage to this model is that it greatly reduces data collision (conflicts between updates to the same record). This is because data changes are made at one site only.

Figure 15–1 illustrates the data distribution model.

Figure 15–1: Data Distribution Model

Consolidation Model

In the consolidation model, data changes are made at the remote sites and then propagated to the central database. The central database is read-only and is used for reporting purposes. For replication, this model increases the frequency of data collision over the distribution model. If there is a collision of changes by two or more users, the changes are applied on a first-come-first-served basis.

To avoid data collision, the consolidation model often uses table partitioning. Table partitioning (also called data ownership) requires that all data be owned by each site. Changes to data at each remote site are made exclusively by respective remote site users. A data ownership model might not be appropriate for your business organization. Although data collisions are avoided, the ability to update the same record from any site is lost.

Figure 15–2 illustrates two data consolidation models, one with no data ownership, and the other with table partitioning.

Figure 15–2: Data Consolidation Models

Peer-to-peer Model

In a peer-to-peer model (or “update anywhere”) any user at any site can update data. This is the most flexible replication model. However, in a peer-to-peer scheme, data collision is a side effect that must be addressed. Data collision must be resolved based on business requirements. An approach to data collision resolution is discussed later in this chapter.

Figure 15–3 illustrates the peer-to-peer model.

Figure 15–3: Peer-to-peer Model


Copyright © 2004 Progress Software Corporation
www.progress.com
Voice: (781) 280-4000
Fax: (781) 280-4095