1. DISTRIBUTED DATABASES JORGE POMBAR

2. Overview Most businesses need to support databases at multiple sites. Most businesses need to support databases at multiple sites. We need a single application that can access multiple databases. We need a single application that can access multiple databases. The goal is for the client to connect to a single server that can issue queries that affect all databases. The goal is for the client to connect to a single server that can issue queries that affect all databases.

3. Problems Some problems come up: Some problems come up: Each application must know and exploit the distribution of data in all of the multiple databases. Each application must know and exploit the distribution of data in all of the multiple databases. Also the DBMS is responsible for maintaining consistency among all databases. Also the DBMS is responsible for maintaining consistency among all databases.

4. Solution We consider this these individual databases to be part of a larger distributed database. We consider this these individual databases to be part of a larger distributed database. Distributed databases are created by allowing servers to interact. Distributed databases are created by allowing servers to interact. We create a server- to-server system. We create a server- to-server system.

5. Principles of Distributed Databases A distributed database is a collection of databases that are related logically but are separated physically. A distributed database is a collection of databases that are related logically but are separated physically. The DBMS needs to be able to use a single database connection to access and modify all of the distributed data. The DBMS needs to be able to use a single database connection to access and modify all of the distributed data. It has a single schema whose tables can be distributed over many different database servers. It has a single schema whose tables can be distributed over many different database servers.

6. Distribution of tables Unfragmented: A table exists on one database and different tables are in different databases. Unfragmented: A table exists on one database and different tables are in different databases. Horizontally fragmented: The rows of the table appear in multiple databases. Each row only appears in one database. Horizontally fragmented: The rows of the table appear in multiple databases. Each row only appears in one database. Vertically fragmented: The columns of a table appear in multiple databases. Only the key columns are duplicated. Vertically fragmented: The columns of a table appear in multiple databases. Only the key columns are duplicated. Replicated: Some or all of the rows and columns are stored in more than one database. Replicated: Some or all of the rows and columns are stored in more than one database. No matter how the tables are distributed, the user must be presented with a schema that makes the distributed database look like a single database.

7. How it works We have a central database server and a local database on each branch. We have a central database server and a local database on each branch. The client application connects to the local database server, which in turn connects to the other database servers in order to access data. The client application connects to the local database server, which in turn connects to the other database servers in order to access data. Any modifications of data requires modification to all connected databases. Any modifications of data requires modification to all connected databases. The client feels that is only interacting with a single, local database. The client feels that is only interacting with a single, local database.

8. Advantages of Distributed Databases Autonomy and availability of data: Each site can always access its local data even if the network connection is down. Autonomy and availability of data: Each site can always access its local data even if the network connection is down. Independence of physical and logical layout: Changes are not the responsibility of client programs. They are done by the database servers. Independence of physical and logical layout: Changes are not the responsibility of client programs. They are done by the database servers. Physical locality: A database application that only needs part of the whole database can access its data locally. So you would get a faster response time. Physical locality: A database application that only needs part of the whole database can access its data locally. So you would get a faster response time. Improved performance: Data is distributed over many computers so query executes faster since we use the processing power of multiple computers. Improved performance: Data is distributed over many computers so query executes faster since we use the processing power of multiple computers.

9. Real world applications Large retail chains: Best Buy, Macys, Circuit City, etc. Large retail chains: Best Buy, Macys, Circuit City, etc. Video rental stores. Video rental stores. Sharing academic research data. Sharing academic research data.

10. References Riccardi, G (2001). Principles of Database Systems with internet and Java applications. Boston, MA: Addison Wesley. Dye, Charles (1999, April). Oracle Distributed Systems. Retrieved May 8, 2008, from O Web site: http://www.oreilly.com/catalog/ordis tsys/chapter1