Shard key selection determines data distribution and performance. Poor shard key choices lead to hotspots, uneven load distribution, and performance degradation across the sharded cluster.
SQL injection occurs when untrusted input is concatenated into SQL queries. Parameterized queries/prepared statements separate code from data, preventing attackers from modifying query logic.
Cloud-native NoSQL databases prioritize availability and partition tolerance (CAP theorem), implementing eventual consistency to handle distributed system challenges while maintaining horizontal scalability.
Data warehouses use materialized views to pre-aggregate data and strategic denormalization for faster analytical queries. These techniques significantly improve performance on large fact tables.
B+ Tree indexes maintain sorted order and support efficient range queries, point queries, and prefix searches. Hash indexes are faster for exact matches but cannot handle range queries efficiently.
NoSQL databases like MongoDB and Cassandra are designed for horizontal scaling and flexible schemas, making them ideal for cloud-based big data applications with varying data structures.
3NF eliminates partial and transitive dependencies, making it the standard for practical database design. BCNF is stricter but 3NF is the most commonly implemented normalization form.
Activation functions (ReLU, sigmoid, tanh) add non-linearity, allowing neural networks to approximate non-linear relationships.
The CAP theorem shows it's impossible to guarantee all three: Consistency, Availability, and Partition tolerance simultaneously.
Consensus algorithms (like Raft, Paxos) enable distributed systems to reliably agree on state even with node failures.