Serviços sem estado versus com estado

When designing systems, a key decision is how your services handle 'state' – information about past interactions.

Services can either remember past client requests (stateful) or treat each request as completely new (stateless).

Understanding this difference is crucial for building systems that can grow and recover from issues.

A stateless service doesn't store any information about previous client requests or sessions. Each request is processed independently.

• Every request contains all the necessary data for the server to fulfill it.
• The server doesn't rely on any stored 'state' from prior interactions.

Think of it like an ATM: you insert your card, complete a transaction, and the machine doesn't 'remember' you for your next visit.

Stateless services offer significant advantages for system scalability and resilience:

• Easy Scaling: You can add or remove servers effortlessly without worrying about losing user-specific data.
• High Resilience: If one server fails, another can immediately take over the next request, as no unique state is tied to the crashed server.
• Simple Load Balancing: Any server can handle any request, simplifying traffic distribution.

Here's a simple Java function. Each call to addNumbers is independent; the function doesn't remember previous sums or user data.

Try running this example:

A stateful service remembers past interactions and stores 'state' from one request to the next.

This state could be user session data, items in a shopping cart, or an ongoing database transaction.

The server needs this stored information to correctly process subsequent requests from the same client.

While necessary in some cases, stateful services introduce complexities:

• Harder to Scale: When adding new servers, you must ensure the correct state is available to the right server, often requiring 'sticky sessions' or shared storage.
• Less Resilient: If a server holding a user's state crashes, that state might be lost, interrupting the user's experience.
• Complex Management: Requires careful design for state synchronization, persistence, and replication across multiple instances.

Despite the challenges, stateful services are sometimes unavoidable or simplify design:

• Databases: They are fundamentally stateful, storing persistent application data.
• User Sessions: Keeping users logged in and tracking their activities across multiple requests.
• Real-time Games: Maintaining the current game state for all connected players.

The key is often to manage state effectively by externalizing it.

Consider a user logging into an e-commerce site. The server needs to remember who they are to show their cart or profile.

This 'session state' could be stored directly on the application server (making it stateful) or, more commonly in scalable systems, in an external, shared state store like Redis or a database.

Externalizing state allows application servers to remain stateless, improving scalability.

• Stateless: No memory of past requests. Each request is complete and independent. Highly scalable and resilient.
• Stateful: Remembers past interactions. Relies on stored context. Harder to scale horizontally and less resilient to individual server failures without complex state management.

The choice impacts how you design for scalability, reliability, and data persistence.

In this lesson, we explored the critical difference between stateless and stateful services:

• Stateless services treat each request independently, offering superior scalability and resilience.
• Stateful services retain information across requests, which can be necessary but introduces scaling and management challenges.

Often, state is moved to external, dedicated stateful stores (like databases or caches) to keep application servers themselves stateless, achieving the best of both worlds.