Request–Reply in NATS: One-to-One Messaging

Pub/Sub vs Request–Reply

Feature	Publish–Subscribe	Request–Reply
Communication style	One-to-many	One-to-one
Response expected	No	Yes
Persistence	Only with JetStream	Optional JetStream, but usually real-time
Typical use cases	Events, notifications	Queries, RPC, commands

Request–Reply vs HTTP Call

At first glance, Request–Reply in NATS looks very similar to making an HTTP request. Both involve a client sending a message and waiting for a response. But there are some clear differences:

Aspect	NATS Request–Reply	HTTP Call
Transport	Uses NATS protocol over TCP (lightweight)	Uses HTTP/HTTPS over TCP (heavier)
Latency	Very low, designed for microseconds	Higher, includes HTTP headers and parsing
Connection	Multiplexed over a single NATS connection	Often opens/closes TCP connections
Routing	Based on subjects (flexible patterns)	Based on URLs and endpoints
Discovery	Implicit — services just subscribe to a subject	Explicit — clients must know exact URL
Scaling	Easy, multiple subscribers can share the load	Usually needs a load balancer or service mesh
Persistence	Works with JetStream if needed	Usually requires external database or logs
Use cases	Fast RPC between microservices, health checks, IoT commands	Web APIs, browser-to-server communication

Example Scenario

• HTTP Call:
  • A mobile app calls GET https://api.example.com/user/123 to fetch user info.
  • The app must know the server’s exact URL and handle authentication, TLS, etc.
• NATS Request–Reply:
  • A service sends a request to subject user.get with payload {id:123}.
  • Any service subscribed to user.get can reply. The client doesn’t care where the service is running.
  • If multiple services are subscribed, NATS will load-balance the request automatically.

Final Thoughts

Request–Reply is a powerful complement to pub/sub in NATS. Use it when you need direct responses from a service, like fetching data, running commands, or performing checks. Together, pub/sub and request–reply give you flexible patterns to design distributed systems that fit your needs.