producer consumer pattern

**Producer-Consumer Pattern** — a fundamental concurrency pattern where producer threads generate data and consumer threads process it, communicating through a shared buffer. **Architecture** ``` [Producer 1] →→ [Producer 2] →→ [Shared Buffer/Queue] →→ [Consumer 1] [Producer 3] →→ →→ [Consumer 2] ``` **Bounded Buffer Implementation** - Fixed-size queue (ring buffer) between producers and consumers - Producers block when buffer is full (back-pressure) - Consumers block when buffer is empty (no work) - Synchronization: Mutex + two condition variables (not_full, not_empty) **Benefits** - **Decoupling**: Producers and consumers run at different speeds - **Buffering**: Absorbs bursts in production/consumption rates - **Scalability**: Add producers or consumers independently **Lock-Free Variants** - **SPSC (Single-Producer Single-Consumer)**: Ring buffer with atomic head/tail pointers — no locks needed. Fastest option when topology matches - **MPMC (Multi-Producer Multi-Consumer)**: More complex, often uses CAS. Examples: Java ConcurrentLinkedQueue, Disruptor **Common Applications** - Web server: Accept thread (producer) → request queue → worker threads (consumers) - Pipeline processing: Each stage is consumer of previous, producer for next - Logging: Application threads produce log entries → log writer consumes **Producer-consumer** is ubiquitous — it appears in virtually every concurrent system from operating systems to web servers.