Databases flashcards, generated for you.

An organized collection of structured data stored and accessed electronically, typically through a DBMS, enabling efficient retrieval, modification, and management.

Atomicity (all-or-nothing), Consistency (valid state), Isolation (no interference), Durability (persistent). Critical for data integrity in transactions, especially financial systems.

Distributed systems can guarantee at most 2 of 3: Consistency (all nodes see same data), Availability (system always responds), Partition tolerance (survives network splits). Real systems choose CP or AP.

Relational: strong consistency, ACID, rigid schema, vertical scaling limits. NoSQL: eventual consistency, flexible schema, horizontal scaling, BASE properties. Choose relational for structured/transactional; NoSQL for scale/flexibility.

Process of organizing data into tables to minimize redundancy and dependency. Benefits: reduces anomalies, saves space. Cost: more joins, slower queries. Normal forms go from 1NF through BCNF.

Deliberately duplicating data across tables to reduce joins and improve query performance. Trade-off: faster reads, slower writes, harder consistency maintenance. Use when reads vastly outnumber writes (analytics, caching).

Data structure (B-tree, hash) enabling fast lookup on indexed columns. Benefit: O(log n) queries vs O(n) table scans. Cost: slower writes, higher storage, maintenance overhead. Index wisely on frequent search/filter columns.

Vertical: add CPU/RAM to single server—simpler, hits hardware limits, no distribution complexity. Horizontal: add more servers—scales indefinitely, requires sharding/replication, introduces consistency/availability complexity and operational burden.

Partitioning data across multiple independent databases by a shard key so each stores a subset. Enables horizontal scaling but introduces: uneven distribution risks, cross-shard join complexity, shard rebalancing difficulty. Choose shard key carefully.

Read replicas: solve read bottlenecks, preserve single-write source, simpler eventual consistency. Sharding: solve write/storage bottlenecks, true horizontal scale, adds query complexity. Use replicas first; shard only when needed.