Overview
Direct Answer
Pub/Sub is an asynchronous messaging architecture in which decoupled applications communicate through intermediary topics, with publishers sending messages to topics and subscribers receiving only those messages they have explicitly subscribed to. This pattern enables many-to-many communication without direct knowledge of recipient endpoints.
How It Works
A message broker or cloud service maintains named topics as central channels. Publishers post messages to these topics without knowledge of subscribers; the broker then routes messages to all subscribed consumers independently. Subscribers connect to topics of interest and receive messages either immediately or from a retained backlog, depending on the system's retention policy and subscription configuration.
Why It Matters
This pattern decouples application components, enabling independent scaling, deployment, and failure isolation—critical for microservices architectures. Asynchronous delivery reduces latency-sensitive dependencies and allows organisations to process high-volume event streams reliably whilst maintaining system resilience during traffic spikes or component failures.
Common Applications
Real-time analytics ingestion, where events from multiple sources flow into a single topic for parallel processing; event-driven notification systems that alert multiple teams; IoT sensor data distribution across analytics, logging, and alerting pipelines; and inventory or order management systems where multiple downstream services react to state changes.
Key Considerations
Exactly-once delivery semantics and message ordering guarantees vary significantly across implementations, requiring careful evaluation against application requirements. Late subscribers may miss historical messages unless retention or replay mechanisms are explicitly configured.
More in Cloud Computing
FinOps
Strategy & EconomicsA cultural practice combining technology, finance, and business to manage cloud costs through data-driven decision making.
AI Infrastructure
Service ModelsThe specialised hardware, software, and networking stack required to train and serve AI models at scale, including GPU clusters, high-bandwidth interconnects, and model serving frameworks.
Cloud Repatriation
Strategy & EconomicsThe process of moving workloads back from public cloud environments to on-premises or private cloud infrastructure.
Load Balancer
InfrastructureA device or software that distributes network traffic across multiple servers to ensure no single server is overwhelmed.
Hybrid Cloud
Strategy & EconomicsAn IT architecture combining on-premises infrastructure with public and private cloud services.
Public Cloud
Service ModelsCloud computing resources shared among multiple organisations and available to the general public over the internet.
Infrastructure as a Service
Service ModelsCloud computing model providing virtualised computing resources like servers, storage, and networking over the internet.
Single Sign-On
Strategy & EconomicsAn authentication scheme allowing users to log in once and gain access to multiple related systems.