A3.4.4 Features of Distributed Databases Notes

What is a Distributed Database

1. A distributed database is a database that is stored across multiple physical locations (servers or cloud).
2. It appears to users as one single database, even though it's spread across sites.

Types of Distributed Databases

1. Homogeneous
   1. All sites use the same DBMS, schema, and OS.
   2. Easier to manage.
2. Heterogeneous
   1. Sites may use different DBMSs, schemas, and OSs.
   2. Requires translation layers and is harder to maintain.

Methods of Distribution

1. Fragmentation
   1. Database is split into parts, and each site manages a fragment.
   2. Useful when local data is accessed more frequently.
2. Replication
   1. Copies of the data (or whole database) are stored at multiple sites.
   2. Improves reliability and availability, but increases complexity.

The Need for Data Consistency

1. Data Consistency ensures that all users see the same data, regardless of which part of the database they access.
2. In a distributed system, data consistency is challenging due to:
   1. Network Latency: Delays in data synchronization across locations.
   2. Concurrent Access: Multiple users updating the same data simultaneously.

The Role of ACID in Distributed Databases

1. ACID stands for Atomicity, Consistency, Isolation, and Durability.
2. These properties ensure reliable transaction processing, even in distributed environments.

Atomicity

1. Atomicity guarantees that a transaction is all-or-nothing.
2. In a distributed database, this means:
   1. If a transaction updates data in multiple locations, either all updates succeed, or none do.

Consistency

1. Consistency ensures that a transaction brings the database from one valid state to another.
2. In distributed systems, this requires:
   1. Synchronizing updates across all locations.

Isolation

1. Isolation ensures that concurrent transactions do not interfere with each other.
2. Techniques like locking and timestamping are used to maintain isolation.

Durability

1. Durability guarantees that once a transaction is committed, it remains so, even in the event of a system failure.
2. This is achieved through replication and backup strategies.

Key Features of Distributed Databases

Concurrency Control

1. Concurrency Control ensures that multiple transactions can occur simultaneously without causing data inconsistencies.
2. Two main approaches:
   1. Pessimistic Concurrency Control (PCC): Locks resources to prevent conflicts.
   2. Optimistic Concurrency Control (OCC): Assumes conflicts are rare and checks for them at commit time.

Data Consistency Models

1. Strong Consistency: Updates are immediately visible across all locations.
2. Eventual Consistency: Updates propagate over time, allowing temporary inconsistencies.
3. Causal Consistency: Preserves the order of causally related updates.

Data Partitioning

1. Data Partitioning divides the database into smaller, manageable sections.
2. Common partitioning methods:
   1. Range Partitioning: Based on value ranges (e.g., dates).
   2. Hash Partitioning: Uses a hash function to distribute data.
   3. List Partitioning: Based on specific values (e.g., regions).
   4. Round-Robin Partitioning: Evenly distributes data across partitions.

Data Security

1. Data Security is critical in distributed systems due to multiple access points.
2. Key security measures:
   1. Authentication: Verifying user identities.
   2. Data Encryption: Protecting data in transit and at rest.
   3. Validated Input: Ensuring data meets predefined rules.

Distribution Transparency

1. Distribution Transparency makes the distributed system appear as a single database to users.
2. Types of transparency:
   1. Access Transparency: Users interact with data the same way, regardless of location.
   2. Location Transparency: Data can be accessed without knowing its physical location.
   3. Replication Transparency: Users are unaware of data copies across sites.
   4. Failure Transparency: The system continues to function despite component failures.
   5. Concurrency Transparency: Multiple users can access data safely.

Fault Tolerance

1. Fault Tolerance ensures the system remains operational despite failures.
2. Achieved through:
   1. Redundancy: Storing multiple copies of data.
   2. Automatic Failover: Switching to backup systems when a failure occurs.

Global Query Processing

1. Global Query Processing optimizes queries across the entire distributed system.
2. Involves:
   1. Local Optimization: Processing queries within individual partitions.
   2. Global Optimization: Combining results from all partitions.

Replication

1. Replication involves storing copies of data at multiple sites.
2. Types of replication:
   1. Full Replication: Entire database is copied to every site.
   2. Partial Replication: Only critical data is replicated.
   3. No Replication: Each site stores unique data.
3. Replication Models:
   1. Master–Slave: One master processes all changes; others follow.
   2. Multi-Master: All nodes can accept updates and sync with others.
   3. Peer-to-Peer: All nodes can act as both master and slave.
   4. Single-Source: One authoritative source replicates to others.

Scalability

1. Scalability allows the system to grow by adding more nodes or partitions.
2. Distributed databases are inherently scalable, making them ideal for large, dynamic environments.

Advantages of Distributed Databases

1. Faster performance through local access
2. Easier to scale by adding sites
3. Improved fault tolerance and reliability
4. Easier data sharing and local autonomy

Disadvantages of Distributed Databases

1. More complex to manage and design
2. Requires strong consistency and concurrency control
3. Harder to secure due to multiple access points
4. Translation issues with heterogeneous systems

Real-World Applications

1. Hotel chains with local hotel data and central reporting
2. Military and air traffic control systems
3. Corporate systems with global departments
4. Pharmaceutical companies tracking treatments globally