Skip to main content
Version: 1.0.1 🏗️

Prefetch Control (Go)

Optimize message delivery by controlling the number of unacknowledged messages.

Basic Prefetch​

easyrmq.NewWorkerBuilder(easyrmq.ExchangeKindDirect).
    WithPool(client).
    WithExchange("order.events.v1").
    Queue("order.process").
    PrefetchCount(10).
    Build(handler)

What is Prefetch?​

Prefetch controls how many messages a consumer can receive at once before acknowledging:

Without Prefetch:          With Prefetch (5):
[Msg1][Msg2][Msg3]...       [Msg1][Msg2][Msg3][Msg4][Msg5]
                              ↑
                              Only 5 unacked messages allowed

Benefits​

✅ Memory Control: Limit memory usage per consumer
✅ Fair Distribution: Spread messages across consumers
✅ Backpressure Handling: Prevent overwhelming consumers

Configuration​

Prefetch Values​

.PrefetchCount(1)   // Process one at a time (strict ordering)
.PrefetchCount(10)  // Balanced performance
.PrefetchCount(50)  // High throughput
.PrefetchCount(0)   // Unlimited (not recommended)

Use Cases​

// CPU-intensive processing
.PrefetchCount(2)

// Fast processing
.PrefetchCount(50)

// Slow I/O operations
.PrefetchCount(5)

// Unpredictable processing time
.PrefetchCount(10)

Best Practices​

  1. Match processing capacity: Set based on your handler's processing speed
  2. Consider resource limits: Account for memory, CPU, and external dependencies
  3. Test and tune: Use load testing to find optimal prefetch values
  4. Multiple queues: Use different prefetch values for different workloads