Database Implementations and N-Tier Architecture (IDS 200 Summer 2025)

Fifteen vocabulary flashcards summarizing major terms, technologies, and concepts from the lecture on modern database architectures and implementations.

N-tier architecture

A database system design separating functionality into 3+ tiers, boosting flexibility, scalability, and maintainability.

Three-tier model

Common form of n-tier architecture with distinct client, application, and database tiers, each running on its own hardware group.

Scalability (in n-tier systems)

The ease of adding machines to individual tiers to handle more users, applications, or data without redesigning the whole system.

Oracle TimesTen

Oracle’s in-memory relational database that keeps data in RAM for speed while persisting changes to disk within an n-tier setup.

Cache model (TimesTen)

Technique where frequently accessed data stays in RAM and less-popular data remains on disk, targeting >90 % of requests served from cache.

MongoDB

Leading NoSQL document database with flexible schema, excellent developer tools, and storage of records in the BSON format.

BSON

Binary JSON; a compact, binary-encoded form of JSON used by MongoDB to reduce storage and transmission size at the cost of extra processing.

Eventual consistency

Consistency model where replicated data copies may diverge temporarily but the system works to make them match over time.

Wide-column store

NoSQL database model whose columns can reference variable-length sets of data, enabling targeted queries on those sets.

Cassandra

Open-source wide-column database originally built at Facebook for high-volume, semi-structured workloads with many small reads and writes.

Haystack

Facebook’s photo storage system that keeps small metadata in RAM and stores image files in an append-only disk database to minimize disk seeks.

Vitess

YouTube-developed middleware that augments MySQL with query parsing, caching, and sharding to support massive traffic and evolving schemas.

Sharding

Splitting a database into independent horizontal or vertical pieces (shards) distributed across servers to boost performance and capacity.

Paxos

Family of consensus algorithms used in Google Spanner to decide correct data versions by requiring a majority of replica nodes to agree on writes.

TrueTime

Google API providing globally synchronized time with bounded uncertainty, enabling Spanner to order transactions consistently across data centers.