Computer Knowledge - MCQ Practice Questions

ACID properties ensure reliable database transactions: Atomicity (all-or-nothing), Consistency (valid state), Isolation (concurrent independence), and Durability (permanent storage).

INNER JOIN returns only rows with matching values in both tables. LEFT JOIN includes unmatched left table rows, FULL OUTER JOIN includes all rows, and CROSS JOIN produces Cartesian product.

LRU cache requires O(1) access and O(1) removal/insertion of least-used items. Hash Map provides O(1) lookup, and Doubly Linked List maintains order efficiently.

Balanced BSTs (AVL, Red-Black) maintain O(log n) search time by keeping height logarithmic. Unbalanced trees degrade to O(n) in worst case.

IaaS provides virtualized computing resources over the internet (AWS EC2, Azure VMs). PaaS and SaaS are higher-level abstractions.

3NF removes transitive dependencies where non-key attributes depend on other non-key attributes. BCNF is stricter but 3NF eliminates transitive dependencies.

Replication maintains copies across nodes for high availability, disaster recovery, and load distribution. It doesn't eliminate backups—they serve different purposes.

Firewalls filter traffic based on security policies. While they provide protection, encryption is separate, antivirus is different, and authentication is handled by other systems.

Modern languages use Introsort which starts with Quicksort (O(n log n) avg) but switches to Heapsort (O(n log n) worst) to avoid worst-case scenarios.

CAP theorem (Brewer's theorem) states distributed systems must choose 2 of 3 properties: strong consistency (C), continuous availability (A), or tolerance to network partitions (P).

Overfitting occurs when a model memorizes training data including noise, reducing generalization. Regularization and cross-validation help prevent it.

NoSQL databases prioritize flexible schemas and horizontal scaling. They may use eventual consistency (BASE model) rather than strict ACID.

Sharding partitions data based on a key (e.g., user ID) across servers. This horizontal scaling improves performance and handles large datasets efficiently.

Activation functions (ReLU, Sigmoid, Tanh) add non-linearity, allowing neural networks to learn non-linear relationships. Without them, networks become linear classifiers.

Microservices decompose applications into loosely-coupled, independently deployable services. This enables scalability, flexibility, and technology diversity.

Connection pooling maintains a pool of reusable connections, reducing the cost of opening/closing connections repeatedly. This significantly improves application performance.

Zero-day vulnerabilities are previously unknown exploits with no patch. They're extremely dangerous because users have no defense until vendors develop and release fixes.

Insertion Sort has O(n) best-case complexity for nearly sorted data. Other algorithms don't exploit this, maintaining O(n log n) regardless of initial order.

BASE is an alternative to ACID, prioritizing availability and partition tolerance. It accepts temporary inconsistencies that resolve eventually (eventual consistency).

Least privilege minimizes security risk by restricting access. Users get only permissions needed for their role, reducing damage from compromised accounts.