System Design Glossary

A single entry point that handles cross-cutting concerns such as auth, rate limiting, and routing for a set of backend services.

A neural network operation that allows each element in a sequence to selectively weight every other element when computing its representation, enabling context-aware processing across arbitrary distances.

Object storage for unstructured binary data: images, videos, documents, ML model weights. Designed for durability and throughput at scale, not low-latency random access.

A space-efficient probabilistic data structure that tests set membership with no false negatives and a tunable false positive rate.

A distributed system can only guarantee two of three: Consistency, Availability, and Partition Tolerance.

Content Delivery Network. A geographically distributed network of edge servers that caches content close to end users, reducing origin server load and cutting time-to-first-byte by 50–200ms depending on user location.

Command Query Responsibility Segregation separates the write model (commands) from the read model (queries) so each can be optimized independently.

A fault tolerance pattern that stops forwarding calls to a failing dependency, giving it time to recover while preventing cascading failures across callers.

A distributed hashing scheme that minimizes key remapping when nodes are added or removed.

Horizontal partitioning of a database across multiple machines to distribute load beyond a single server's capacity.

A shared cache layer across multiple nodes used to absorb read traffic from the primary database and reduce latency on hot data paths. The difference between a 2ms and a 200ms read at scale.

The problem of getting distributed nodes to agree on a single value despite network failures and partial outages. The theoretical foundation behind etcd, ZooKeeper, and Kafka leader election.

Numerical vector representations of text, images, or other data that encode semantic meaning. The translation layer that converts unstructured content into a form that can be compared mathematically.

An architectural pattern where application state is derived by replaying an append-only log of events rather than storing and mutating current state directly.

Log-Structured Merge Tree is a storage structure that converts random writes to sequential writes by buffering mutations in memory before flushing sorted files to disk.

A protocol by which distributed nodes agree on a single authoritative node to coordinate a task, preventing split-brain and duplicate operations.

Distributes incoming traffic across multiple servers to prevent any single node from becoming a bottleneck. The mechanism that makes horizontal scaling functional in practice.

Asynchronous communication buffer between services. Decouples producers from consumers and provides durability during traffic spikes.

Retrieval-Augmented Generation is an architecture that grounds LLM responses in specific retrieved documents, reducing hallucination on facts outside the model training set.

Controls the rate of requests to a service. Common algorithms: Token Bucket, Leaky Bucket, Fixed Window, and Sliding Window.

A pattern for managing distributed transactions as a sequence of local transactions, each with a compensating transaction for rollback when a step fails.

An infrastructure layer that manages service-to-service communication through sidecar proxies, providing retries, mTLS, and observability without application code changes.

A distributed coordination protocol that ensures all participants in a transaction either commit or abort atomically, using a prepare phase followed by a commit phase.

A database purpose-built to store and query high-dimensional embedding vectors. The retrieval layer that makes semantic search and RAG pipelines possible at production scale.