Resources · Reference

Glossary

Concise definitions for the key terms, acronyms, and concepts used throughout this library — organized alphabetically.

ACID: Atomicity, Consistency, Isolation, Durability — the four properties that guarantee reliable database transactions. Contrast with BASE.
ADR (Architecture Decision Record): A short document capturing the context, decision, and consequences of a significant architectural choice. See ADR Template.
Anti-Corruption Layer (ACL): A translation boundary that isolates a domain model from external system models, preventing external terminology and concepts from leaking inward.
Availability: The proportion of time a system is operational and accessible. Often expressed as nines: 99.9% = 8.7h downtime/year, 99.99% = 52min/year.

BASE: Basically Available, Soft state, Eventually consistent — a relaxed consistency model used by many NoSQL systems. Contrast with ACID.
Bounded Context: A logical boundary within which a domain model is internally consistent and terms have unambiguous meaning. A core concept in Domain-Driven Design.
Bulkhead: A resilience pattern that isolates resource pools to prevent failures in one area from cascading to another. Named after ship compartments.

CAP Theorem: In a distributed system, you can only guarantee two of: Consistency, Availability, Partition tolerance. See CAP Theorem.
CDC (Change Data Capture): Tracking and publishing database changes as events, typically via log-based mechanisms like Debezium.
Circuit Breaker: A pattern that trips open when a downstream service fails repeatedly, stopping calls temporarily to allow recovery and prevent cascading failures.
CQRS: Command Query Responsibility Segregation — separating the write (command) and read (query) sides of a domain model.

Dead Letter Queue (DLQ): A queue that receives messages that could not be processed successfully, for later inspection and reprocessing.
Durability: The guarantee that committed data persists even after system failures. One of the four ACID properties.
Domain-Driven Design (DDD): A software design approach that aligns code structure with business domains using aggregates, entities, value objects, and bounded contexts.

Error Budget: The allowed amount of downtime or unreliability before an SLO is breached, calculated as 1 minus the SLO target over a rolling window.
Eventual Consistency: A consistency model where replicas will converge to the same state given sufficient time without new updates.
Event Sourcing: A pattern where application state is derived from an immutable sequence of events rather than from current-state snapshots.

F–I

Fanout: Distributing one message or event to multiple consumers or queues simultaneously.
Idempotency: The property that an operation can be applied multiple times without changing the result beyond the first application. Critical for safe retries.
Ingress: In Kubernetes, a resource that manages external access to services, typically HTTP/HTTPS routing with TLS termination.

L–N

Latency: The time elapsed between a request being issued and the response beginning to arrive. Typically measured as p50, p95, p99 percentiles.
Leader Election: A distributed algorithm for selecting one node from a group to act as coordinator or primary for a set of operations.
Microservices: An architectural style where an application is composed of small, independently deployable services, each owning its own data and business capability.
MTTR / MTBF: Mean Time To Recovery and Mean Time Between Failures — key reliability metrics measuring how fast you recover and how often failures occur.

O–P

Observability: The ability to understand a system's internal state from its external outputs — logs, metrics, and traces are the three pillars.
PACELC Theorem: Extension of CAP: even when a system is available (no partition), there is a trade-off between latency and consistency.
Partition Tolerance: A system's ability to continue operating when network partitions split it into disconnected groups.
Pub/Sub: A messaging pattern where publishers send events to topics without knowledge of subscribers, and subscribers receive events independently.

Q–R

Quorum: A majority of nodes in a distributed system agreeing on a value or action to ensure consistency despite failures.
Rate Limiting: Controlling the rate at which requests are accepted from a client or system to protect capacity and ensure fairness.
Read Replica: A database instance that receives a stream of changes from the primary and serves read-only queries to reduce primary load.
RTO / RPO: Recovery Time Objective (max acceptable downtime) and Recovery Point Objective (max acceptable data loss) — disaster recovery targets.

Saga: A distributed transaction pattern using a sequence of local transactions with compensating actions for rollback on failure.
Service Mesh: Infrastructure layer handling service-to-service communication with traffic management, mutual TLS, and observability.
Sharding: Horizontally partitioning data across multiple nodes, each holding a subset (shard) of the total data.
SLI / SLO / SLA: Service Level Indicator (the metric), Service Level Objective (the target), and Service Level Agreement (the contract).
Strangler Fig: A migration pattern for incrementally replacing a legacy system by routing traffic to a new system alongside the old one.

T–Z

Throttling: Temporarily reducing the rate of incoming requests to a service, typically with back-pressure or queuing, to avoid overload.
Two-Phase Commit (2PC): A distributed protocol where a coordinator asks all participants to prepare a transaction, then commits if all are ready.
Ubiquitous Language: In DDD, a shared vocabulary between domain experts and developers that is reflected consistently in code, conversations, and documentation.
Zero Trust: A security model that eliminates implicit trust based on network location — every request is authenticated and authorized regardless of origin.