Distributed Database Design–Factors of DDBMS

The methodology used for the logical design of a centralized database applies to the design of a distributed one as well. However, for a distributed database design three additional factors have to be considered:

Data Fragmentation

The database may be broken up into logical units called fragments. These fragments are stored at different sites. Data fragmentation is the design process of deciding what the fragments will be. The simplest logical units are the relations themselves. Each relation is stored at a particular site. In many cases, it makes sense to divide a relation into smaller logical units for distribution.

Horizontal & Vertical Fragmentation

A vertical partition, or vertical fragment, refers to a table that is broken into two or more sets of columns. A horizontal partition, or horizontal fragment, refers to the rows of a table when they are divided into pieces.

Data Replication

A copy of each fragment may be stored at several sites. Data replication is the design process of deciding which fragments will be replicated.

Data Allocation

Each fragment has to be allocated to one or more sites, where it will be stored. There are three strategies regarding the allocation of the data:

Fragmented Data Allocation

The database is partitioned into disjoint fragments. Each fragment is assigned to one site and there is no replication. This is also called non-redundant allocation.

Advantages

Some advantages of this data allocation method are as follows:

• Efficiency: Data are stored close to where they are used and separate from non-local data. So it can be accessed and used efficiently.

• Security: Data that is not relevant at a particular site is not made available. It provides more security.

Disadvantages

Some disadvantages of this data allocation method are as follows:

• Inconsistent access speeds: Access to remote fragments takes much longer than to the local one.

Back up Vulnerability: Since data are not replicated, problems at a site make fragments inaccessible.

Complete Replication

A complete copy of the database is maintained at each site and there is no fragmentation

Advantages

Some advantages of this data allocation method are as follows:

• Faster & Consistent Access Speeds: Each site that has a full copy can process queries locally, so queries can be processed rapidly.

• Reliability: If one of the sites containing a relation (or the database) fails, a copy can always be found at another site. The available copies can all be updated as soon as possible as transactions occur. The unavailable nodes can be updated.

Node Decoupling: Each transaction may proceed without coordination across the network. So, if nodes are down, busy, or disconnected, a transaction is handled when the user desires.

Disadvantages

Some disadvantages of this data allocation method are as follows:

• Update complexity: Whenever there is an update to one of the database copies, the update has to be applied to all the other copies.

• Increased Storage Requirements: Each site that has a full copy must have the same storage capacity that would be required if data were stored centrally Each copy requires storage space and the processing time is required to update each copy on each node.

Selective Replication

Selective replication is a combination of fragmentation and replication. The objective of this strategy is to have the advantages of the previous approaches without the disadvantages.