JAWAHARLAL NEHRU TECHNOLOGICAL UNIVERSITY HYDERABAD MCA DISTRIBUTED DATABASES Previous Question Paper (September/October - 2022 - R15)

A centralized database stores all data at a single physical location, managed by a central server. Access is controlled by one site, making it easier to maintain consistency and security, but it creates a single point of failure and potential performance bottlenecks as the number of users grows.

A distributed database consists of data stored across multiple physically dispersed sites connected by a network. It offers high availability, fault tolerance, and improved performance by allowing local access to data. However, it introduces significant complexity in ensuring data consistency, managing distributed transactions, and handling network communication overhead.

Distributed database design involves determining how data and applications are distributed across multiple sites to optimize performance, reliability, and availability. The two primary strategies are:

1. Fragmentation: The process of dividing a global relation into smaller logical units called fragments. This can be done via horizontal fragmentation (subset of rows), vertical fragmentation (subset of columns), or hybrid fragmentation. Fragmentation allows data to be stored closer to where it is most frequently accessed.
2. Allocation (Replication): Deciding where to store each fragment. Fragments can be stored at a single site or replicated across multiple sites. Replication increases data availability and read performance but complicates update operations, as all copies must be synchronized to maintain consistency.

The design process aims to minimize communication costs, maximize local processing, and ensure that the system meets specific performance requirements while maintaining data integrity across the network.