High-Performance RPC: Microservices with gRPC & Protocol Buffers

Introduction: service-to-service bottlenecks

gRPC runs over HTTP/2 with compact Protocol Buffers payloads—strong schemas, efficient binary encoding, and first-class streaming.

Covers: .proto basics, protoc codegen, synchronous server/client sketches, streaming modes, errors/metadata, performance tips, production checklist.

---

Table of contents

1. Protocol Buffers basics
2. gRPC C++ server & client
3. Streaming
4. Complete examples
5. Common errors
6. Performance
7. Production patterns
8. Checklist
9. Summary

---

1. Protocol Buffers

Define messages/services in .proto, run protoc with grpc_cpp_plugin to generate *.pb.{h,cc} and *.grpc.pb.{h,cc}. Never reuse field numbers for different meanings—compatibility hinges on field IDs.

syntax = "proto3";
package myapp;

message UserRequest { int32 user_id = 1; string name = 2; }
message UserResponse { int32 user_id = 1; string name = 2; int64 created_at = 3; }

service UserService {
  rpc GetUser(UserRequest) returns (UserResponse);
}

---

2. gRPC C++

ServerBuilder + service implementation; client uses CreateChannel + stub. Always check status.ok() and set deadlines on ClientContext.

---

3. Streaming

Server streaming, client streaming, bidi—choose based on data direction; handle Write failures and IsCancelled() on long RPCs.

---

4. Complete examples

Add metadata/auth snippets, retry with backoff, and TLS credential setup in production services—the patterns are standard gRPC client configuration.

---

5. Common errors

UNAVAILABLE/Connection refused → server down or wrong address; DEADLINE_EXCEEDED → increase deadline or optimize server; RESOURCE_EXHAUSTED → raise message size limits via ChannelArguments.

---

6. Performance

Reuse channels/stubs; prefer bytes over giant JSON strings inside protobuf; reserve repeated fields; HTTP/2 multiplexes many RPCs per connection.

---

7. Production patterns

Health checks, metadata propagation (x-request-id), graceful Shutdown(), TLS for untrusted networks, schema evolution policy.

---

8. Checklist

• Deadlines on clients
• Status codes meaningful
• TLS where required
• Stream cancellation handling

---

9. Summary

Protobuf + gRPC gives typed, fast service RPC with streaming; operate it with observability and clear compatibility rules.

Next: Secure coding & OpenSSL (#43-2)
Previous: Linux syscalls (#42-3)

Keywords

gRPC, Protocol Buffers, C++ microservices, HTTP/2 RPC, protobuf codegen