Transitioning from Monolith to Microservices

Overview

• Moving from a monolithic application to a cloud-based microservices architecture involves several challenges.
• In a monolith, all functionalities are on the same server.
• Example: Updating inventory after an order is a local call in a monolith.
• In microservices, services can be dynamic, requiring routing and registration methods.

Service Discovery

• Definition: Process of locating resources (applications/services) on a network.
• Requires a Service Registry: a database storing services and their locations.
• Example of Service Registry: DNS (Domain Name Service) which maps IP addresses to domain names.
• Importance of keeping the Service Registry up to date:
  • Remove services that shut down.
  • Add new services as they start.
  • Remove unhealthy services from the registry.

Types of Service Discovery

1. Where is Service Discovery Happening?

• Client-Side Service Discovery:
  • Service A queries the Service Registry for the location of Service B and communicates directly.
  • Example: Similar to how a browser uses DNS to find a website.
• Server-Side Service Discovery:
  • Service A communicates through a Proxy that queries the Service Registry and redirects the request to Service B.

2. How is the Service Registry Maintained?

• Self Registration:
  • Each service updates the Service Registry with its status.
  • Pros: Fine-grained status updates.
  • Cons: Services hanging or not responding can't update the registry; requires specific implementations for each language.
• Third-party Registration:
  • An external application (Registrar) observes and updates the Service Registry.
  • Types: Sidecar or Service Orchestration Engine (e.g., Kubernetes).
  • Pros: Language agnostic, can check unresponsive services.
  • Cons: Requires additional infrastructure or limited to services deployed within the orchestration engine.

Pros and Cons Summary

Client-Side Service Discovery

• Pros:
  • Simple implementation.
• Cons:
  • Requires specific implementations for each programming language.
  • More coupling with Service Registry.

Server-Side Service Discovery

• Pros:
  • Language agnostic; no direct coupling with Service Registry.
• Cons:
  • Requires infrastructure setup.

Self Registration

• Pros:
  • Detailed status awareness.
• Cons:
  • Potential outdated registry for unresponsive services.
  • Language-specific implementations required.

Third-party Registration

• Pros:
  • Language agnostic, external checks for service status.
• Cons:
  • Dependency on a specific registration system.
  • Limited to services within the orchestration engine.

Open Questions & Future Topics

• Challenges of deploying services outside of the main infrastructure.
• Handling performance issues and traffic routing intelligently.
• Introduction of Service Mesh as a solution for more complex routing and performance management.

Conclusion

• Various Service Discovery approaches exist, each with benefits and drawbacks.
• Performance and smart traffic routing are advanced topics to explore further.