Key Differences Between RPC and Message Queues (MQ) in Distributed Systems

RPC (Remote Procedure Call)

RPC is a remote procedure call mechanism that abstracts away the underlying network communication, allowing a client to invoke methods on a remote server as if they were local.

Common RPC frameworks include gRPC (Google), Thrift (Facebook), and Dubbo (Alibaba).

Typical RPC workflow (four steps):

1. Establish Communication : Create a TCP connection between client and server.

2. Service Addressing : Determine the target host/IP, port, and method name.

3. Network Transmission : Serialization – convert parameters to a binary format. Deserialization – reconstruct parameters on the receiving side.

4. Service Invocation : The server processes the request (often via a proxy) and returns a serialized response, which is transmitted back to the client.

These four steps constitute a complete RPC call.

MQ (Message Queue)

Message queues provide a communication model where producers send messages to a queue that consumers later retrieve, enabling asynchronous or synchronous processing.

Typical MQ middleware includes RabbitMQ, RocketMQ, and Kafka.

Key characteristics of MQ:

Decoupling of producers and consumers

Reliable delivery

Broadcast capabilities

Eventual consistency

Traffic shaping (peak‑shaving)

Message delivery guarantees

Support for both asynchronous and synchronous communication

Improved system throughput and robustness

Typical use case: using MQ for traffic shaping in flash‑sale (秒杀) scenarios and for decoupling services.

Four Major Differences Between RPC and MQ

1. Architectural Difference : MQ introduces an intermediate message queue that stores messages, whereas RPC communicates directly between client and server.

2. Invocation Difference : For immediate, synchronous responses, RPC is preferred.

3. Functional Difference : MQ focuses on buffering and decoupling, supporting patterns like point‑to‑point and publish‑subscribe, whereas RPC is a direct call mechanism.

4. Performance Difference : When a service becomes a bottleneck, converting synchronous RPC calls to asynchronous MQ processing can improve scalability and throughput.

For more detailed discussions on distributed architecture, Dubbo, and micro‑service design, refer to the linked articles in the original source.

This series of collections covers distributed systems, middleware, micro‑services, flash‑sale architectures, and large‑scale website design, helping professionals advance their high‑salary career paths.