Distributed Database Concepts

A distributed database should have the following desired attributes¹

• Distributed data independence, users should be able to ask queries without specifying the referenced relations
• Distributed transaction atomicity, users should be able to update and access data across sites

When a slow connection is used between sites these attributes may not be achievable or desirable.

Data storage across multiple databases can use one or both of the following techniques²

• Fragmentation
• Replication

to increase speed and responsiveness.

Fragmentation may use

• Horizontal or
• Vertical fragmentation

Replication allows

• Increased availability of data
• Faster query evaluation

The current approach we are considering is fragmentation paired with replication. We will fragment the data to manage ownership of data to avoid the sync problems associated with multiple changes to the same record of data from multiple sites. Read only queries will be executed on the local replicated data because this will allow better data availability and faster query evaluation.

References:

1. p736 Database Management Systems, Ramakrishnan
2. p739-p741 Database Management Systems, Ramakrishnan