Why RPC Is the Backbone of Modern Distributed Systems

RPC, one of the mainstream technologies today, breaks the old notion that a single task must run on a single computer, influencing distributed computing and micro‑services.

Since the 1980s, many middleware technologies have evolved from RPC, thanks to advances in networking and inter‑process communication.

An RPC framework is a reusable software architecture that implements RPC; understanding it requires knowing how to use the framework, its abstractions, and the service‑governance features it provides.

Remote Procedure Call (RPC) means invoking a service on a remote machine over a network, effectively placing part of a program on another host and returning the result.

Remote procedure : a piece of logic deployed on another machine, enabling business decomposition, independent scalability, and easier maintenance.

Procedure call : the familiar function or method invocation, which when combined with a remote procedure allows cross‑machine execution and data transfer.

For example, an order‑creation service on machine A and a payment service on machine B are invoked sequentially by an ordering service, illustrating RPC.

RPC can be likened to a long‑distance relationship: two services communicate via a protocol (the “secret language”), and both must agree on message encoding and decoding.

Serialization must be consistent, otherwise calls fail; detailed components are covered in later chapters of “Deep Understanding of RPC Framework Principles and Implementation”.

Early monolithic applications placed all functions in one process, but growing user bases and complexity made single‑application architectures inefficient, leading to vertical (service‑oriented) architectures.

Vertical architectures separate concerns (e.g., merchant management vs. buyer shopping) to improve scalability and maintainability, but they introduce inter‑service interactions that rely on RPC.

While RPC enables distributed systems, it also brings challenges:

Communication latency and encoding/decoding overhead.

Address‑space isolation, making pointers meaningless across machines.

Partial failures requiring service discovery and fault‑tolerance mechanisms.

Concurrency issues when many nodes invoke the same service simultaneously.

These challenges have motivated solutions such as registration centers, distributed transactions, and improved network protocols.

The discussion is excerpted from the book “Deep Understanding of RPC Framework Principles and Implementation”, which also provides implementation examples and selection guidelines.