Designing the Data Structure › Choose to Distribute or Replicate Data › Challenges of Distributed and Replicated Data › Integrity of Distributed Data

Integrity of Distributed Data

If the data is not distributed and held on a single server, data integrity raises the same issues as in a traditional host/terminal environment. However, if the data is to be distributed across two or more processors, you will need to consider how the data is to be distributed.

Distributing data has a major consequence on the integrity of the data. Take special care in designing data distribution. The integrity of the data to be stored within the database must be guaranteed. It is vital that the database management system to be used fully supports the use of distributed data. In particular, the database management system needs to handle the checking and cancellation of unsuccessful changes to avoid integrity problems when updating the data.

The following list explains options for distributed data:

• Rows in tables may be arranged by value. For example, all rows for customers of each sales office could be placed contiguously.
• Updating the data is relatively straightforward. Each processor can update the data independently of the others.
• Columns in tables may be split by ownership. For example, the sales department has CUSTOMER NAME and ADDRESS, and the accounts department has CUSTOMER RATING.
• In this case, the data at one location should be maintained as the master. Data at any other location that uses or updates that data should be kept in synchronization with that master.
• A combination of splitting by rows and tables may be used.

Deletion of any data in a distributed environment can cause problems, as other processors may be dependent on the data for their own procedures. Deletions need to be synchronized, possibly by performing such operations at predetermined times, such as overnight.