Questions
1–2 questions in major university papers
Difficulty
Medium
Importance
High yield for DBMS theory exams and Viva
Overview
Distributed databases extend storage and processing across multiple nodes to improve availability and scalability. Mastery of this topic is essential for understanding modern cloud architectures and the limitations imposed by network partitions, which are frequent subjects in university theory exams.
NoSQL Database Models
NoSQL databases provide flexible schemas for unstructured data, categorized by their primary data storage model. These are designed for horizontal scalability, contrasting sharply with the rigid relational models of traditional RDBMS.
- Key-Value Stores: Simplest model using a hash table (e.g., Redis, DynamoDB)
- Document Stores: JSON/BSON format for hierarchical data (e.g., MongoDB)
- Columnar Stores: Optimized for analytical queries on large datasets (e.g., Cassandra)
- Graph Databases: Stores entities and their relationships (e.g., Neo4j)
- Schema-less architecture allows dynamic field addition
The CAP Theorem
The CAP theorem states that a distributed system can only provide two of three guarantees: Consistency, Availability, and Partition Tolerance. In the presence of a network failure (Partition), architects must choose between system availability and data consistency.
- Consistency: Every read receives the most recent write
- Availability: Every request receives a non-error response
- Partition Tolerance: System functions despite message loss between nodes
- PACELC Theorem: Extension addressing latency vs consistency during normal operation
- CP systems prioritize data accuracy over uptime
- AP systems prioritize continuous access over immediate consistency
Query Processing and Join Strategies
Query optimization in distributed databases focuses on minimizing network communication costs, which is the primary performance bottleneck. Strategies involve moving data to the query site or query code to the data site to reduce bandwidth consumption.
- Semi-join: Reduces volume of data transmitted by filtering tuples before transfer
- Bloom Join: Uses hash filters to reduce network I/O
- Global Query Optimizer: Transforms algebraic expressions to minimize total cost
- Fragmentation transparency: Hiding the fact that data is stored in fragments
- Location transparency: User query remains agnostic of data physical location
Formula Sheet
Total Cost = Local Processing Cost + Communication Cost
Semi-join(R, S) = Project(R, JoinKeys) |X| S
Exam Tip
Always draw the CAP triangle diagram in your exam script, as it acts as a visual anchor that examiners look for to award full marks.
Common Mistakes
- Confusing the CAP theorem by assuming a system can achieve all three guarantees simultaneously under all conditions.
- Neglecting the cost of network data transmission when explaining distributed join algorithms.
- Failing to distinguish between horizontal scaling (NoSQL) and vertical scaling (Relational).
More Revision Notes
Ready to test yourself?
Play topic-wise Distributed Databases & Query Optimization questions in Aspirant Arcade — gamified MCQ practice.
Download Free