DBMS:Centralised vs Distributed

Large commercial databases may exist in two different Topologies. In both cases the database "looks" like one database.
 * Centralised - where the database is physically in one location and users typically use an Internet connection to access it. Banks (such as ANZ) tend to use centralised databases.
 * Distributed - Where the database is in many locations often where you have a national or international company and customers tend to regularly interact with a local branch. For example: Google uses Big-Table a distributed DBMS as searching tends to be by users in a particular region of the world.

Centralised

 * A single database maintained in one location.
 * Managed by a database administrator.  (usually )
 * Access via a communications network
 * LAN
 * WAN
 * Terminals provide distributed access

Examples

 * Some major banks do all their processing on a mainframe, in some cases in a different country.
 * Clients may use several branches, and online banking for transactions.
 * Airline reservation systems need to be centralised to avoid double bookings.
 * Inland Revenue in New Zealand is countrywide
 * In NZ Police and ambulance calls are sent to a central call center.

Advantages

 * Increased reliability and availability
 * Modular (incremental) growth
 * Lower communication costs
 * Faster Response

Disadvantages

 * Software cost and complexity
 * Processing overheads
 * Data integrity

Distributed database
A single logical database that is spread physically across computers in multiple locations that are connected by a data communications link.
 * Most processing is local
 * Need for local ownership of data
 * Data sharing require

Note that users think they are working with a single corporate database

Examples

 * Chain Stores like the MSD Spears (50% locally owned)
 * Google: Use a DBMS called Bigtable. (Note it is not a Relational Database).("What database does Google use?", 2010 ; Chang,F., et al.,2006 )

Advantages

 * Minimise communications
 * Costs
 * Local control

Disadvantages

 * Adds to complexity and cost
 * Processing overheads
 * Data Integrity