Concurrency Patterns

The Master-Slave pattern is a concurrency model where one thread (the master) distributes work to multiple other threads (slaves). The master thread typically manages the tasks, assigns them to available slaves, and aggregates the results. Slaves operate independently, processing their assigned tasks without direct …

Read More

The Half-Sync/Half-Async pattern addresses performance bottlenecks in request processing by decoupling time-consuming, non-essential tasks from the critical path. It allows a service to respond quickly to a client by handling essential parts of a request synchronously, then offloading less important, potentially …

Read More

The Scheduler-Agent-Supervisor pattern addresses the challenge of reliably executing tasks in a distributed or concurrent environment. A Scheduler is responsible for creating and assigning tasks to Agents. Agents execute these tasks and report their status. A Supervisor independently monitors the Agents and, if an …

Read More

The Monitor Object pattern provides a mechanism to control access to a shared resource in a concurrent environment. It encapsulates the shared resource and its associated access methods, ensuring that only one thread can operate on the resource at any given time. This is achieved through the use of internal locking …

Read More

The Guarded Suspension pattern addresses the problem of a thread needing to wait for a resource that isn’t immediately available. Instead of busy-waiting (repeatedly checking for availability), the requesting thread suspends itself, and a separate “guard” thread monitors the resource. When the …

Read More

The Reactor pattern decouples event handling from the actual event sources. An event loop (“reactor”) monitors multiple event sources (like network sockets, timers, or user input) and dispatches events to associated handler functions when they occur. This allows for concurrent handling of multiple events within a …

Read More

The Message Channel pattern provides a way for components of a system to communicate with each other without direct references. It introduces an intermediary channel through which messages are sent and received, decoupling the sender (client) and receiver (server). This is particularly useful in concurrent or …

Read More

The Leader-Follower pattern is a concurrency pattern used to maintain data consistency across multiple nodes in a distributed system. One node is designated as the “leader” and is responsible for handling all write operations. Other nodes are “followers” that replicate the leader’s data …

Read More

The Proactor pattern is a concurrent design pattern that tackles the challenges of handling numerous asynchronous operations. It decouples the initiation of an asynchronous operation from its completion, allowing a single thread (the proactor) to manage multiple operations without blocking. When an operation completes, …

Read More

The Supervisor-Worker pattern addresses the challenges of managing and maintaining long-running processes or tasks. A Supervisor component is responsible for monitoring and controlling one or more Worker components. The Workers perform the actual work, while the Supervisor ensures that Workers stay alive, restarts …

Read More