715 all vocab

ACID

Properties that ensure reliable database transactions: Atomicity, Consistency, Isolation, Durability.

Atomicity

A transaction happens completely or not at all. ALL OR NOTHING

Consistency

A transaction moves the database from one valid state to another.

Isolation

Transactions do not interfere with one another.

Durability

Committed data remains even after a crash.

CAP Theorem

A distributed system can guarantee only two of the following three:

• Consistency

• Availability

• Partition Tolerance

OLTP

Processes many small, real-time transactions.

OLAP

Analyzes large amounts of historical data for reporting and decision-making.

SQL Database

A relational database that stores data in tables with fixed schemas.

NoSQL Database

A non-relational database designed for flexible, scalable storage of structured or unstructured data.

Throughput

How much work gets done

Latency

Waiting time. Time required to get a response

Horizontal Scaling

Adds more servers to increase capacity. (growth is faster/ foundation of distributed systems and NoSQL)

Vertical Scaling

Adds more CPU, RAM, or storage to one server.

Replication

Creates copies of data on multiple servers for reliability and availability.

Sharding

Splits data across multiple servers to improve scalability.

Hadoop

A framework for storing and processing large datasets across many computers.

HDFS

Hadoop Distributed File System for storing large datasets across multiple machines

Map

Definition:
Processes input data into key-value pairs.

Reduce

Combines intermediate results into a final output.

Spark

A distributed computing framework that processes data primarily in memory for high speed.

ETL

Extract, Transform, Load.

Data is transformed before loading.

ELT

Extract, Load, Transform.

Data is transformed after loading.

Data Warehouse

Stores structured, cleaned data for analytics.

Data Lake

Stores raw structured and unstructured data.

Graph

A data structure made of nodes connected by relationships.

Node

An entity in a graph.

Relationship (Edge)

A connection between two nodes.

Property

Information stored on a node or relationship.

Neo4j

A native graph database.

Cypher

Neo4j's query language.

Index-Free Adjacency

Nodes directly reference neighboring nodes, enabling fast graph traversal

BFS (Breadth-First Search)

A graph traversal algorithm that explores level by level using a queue and finds shortest paths in unweighted graphs.

DFS (Depth-First Search)

A graph traversal algorithm that explores as deep as possible using a stack (or recursion).

Dijkstra's Algorithm

Finds the shortest path in a weighted graph with non-negative edge weights using a priority queue.

Queue

FIFO (First-In, First-Out) data structure

Stack

LIFO (Last-In, First-Out) data structure.

Priority Queue

A queue that removes the item with the highest priority (lowest distance in Dijkstra).

Data Stream

A continuous, real-time, potentially infinite flow of data.

Event-Time

The time an event actually occurred.

Processing-Time

The time the system processes an event.

Watermark

A mechanism that tracks stream progress and helps handle late events.

Tumbling Window

A non-overlapping fixed-size window.

Sliding Window

An overlapping fixed-size window.

Session Window

A window based on user activity that closes after inactivity.

Record-by-Record Processing

Processes each event immediately as it arrives.

Micro-Batching

Processes small batches of events at regular intervals.

Stateful Processing

Maintains information across multiple events.

CEP (Complex Event Processing)

Detects meaningful patterns across multiple events.

Lambda Architecture

A three-layer architecture combining batch and stream processing.

Kappa Architecture

A single stream-processing architecture that replays event logs

IoT Pipeline

Processes IoT data through collection, preprocessing, processing, and visualization.

IaaS

Infrastructure as a Service; provides virtual hardware while users manage the operating system and applications.

PaaS

Platform as a Service; users deploy applications while the provider manages the platform.

SaaS

Software as a Service; complete software provided over the internet.

FaaS (Serverless)

Function as a Service; executes individual functions on demand without managing servers.

Stateless

Does not retain information between executions.

Ephemeral

Temporary; exists only while running.

Auto-Scaling

Automatically adjusts computing resources based on demand.

Cold Start

Initialization delay when an inactive serverless function runs again.

Storage Disaggregation

Separates compute resources from storage so each can scale independently.

LLM

A Large Language Model trained on massive text datasets to generate human-like text.

Transformer

The neural network architecture used by modern LLMs.

GPT

Generative Pre-trained Transformer; predicts the next token.

Token

A piece of text processed by an LLM.

Embedding

A numerical vector representation of data.

Attention

A mechanism that focuses on relevant surrounding tokens to understand context.

Vector Database

A database that stores and searches embeddings for similarity search.

Semantic Search

Search based on meaning instead of exact keywords.

Cosine Similarity

Measures similarity using the angle between vectors.

Euclidean Distance

Measures straight-line distance between vectors.

KNN

Exact nearest-neighbor search that compares against every vector.

ANN

Approximate nearest-neighbor search that is faster but slightly less accurate.

HNSW

An Approximate Nearest Neighbor (ANN) search algorithm.

Chroma

ANN open-source vector database.

Pinecone

A cloud-native vector database

RAG (Retrieval-Augmented Generation)

A technique that retrieves relevant documents from a vector database before an LLM generates an answer, improving accuracy and reducing hallucinations.

Hallucination

An incorrect or unsupported answer generated by an LLM.