Asynchronous communication — queues, topics and events — Microservices & System Design with .NET

4. Asynchronous communication — queues, topics and events

Intermediate★ Must knowLesson 4 of 9

Concept	Meaning
Queue (point-to-point)	One message → one consumer (work distribution). Service Bus queue, RabbitMQ queue
Topic / pub-sub	One message → every subscriber gets a copy. Service Bus topics, RabbitMQ fanout exchange, Kafka
Command vs event	Command = "do this" (one handler). Event = "this happened" (zero or many listeners)
At-least-once delivery	A message can arrive twice → consumers must be idempotent
Dead-letter queue	Where messages go after repeated failures, for inspection and replay
Event streaming	Ordered, replayable log (Kafka, Event Hubs) — for analytics, IoT, event sourcing

csharp

// MassTransit consumer — idempotent by checking a processed-message table
public class OrderPlacedConsumer(AppDb db) : IConsumer<OrderPlaced>
{
    public async Task Consume(ConsumeContext<OrderPlaced> ctx)
    {
        if (await db.Processed.AnyAsync(p => p.MessageId == ctx.MessageId)) return;   // already handled
        db.Invoices.Add(new Invoice(ctx.Message.OrderId, ctx.Message.Total));
        db.Processed.Add(new ProcessedMessage(ctx.MessageId!.Value));
        await db.SaveChangesAsync();
    }
}

Benefits: temporal decoupling (consumer can be down), load levelling (queue absorbs peaks), easy to add new subscribers.