อธิบาย Saga แบบมีตัวประสานงาน

Welcome to the Orchestration Saga! In a distributed system, sometimes a business transaction needs to involve multiple services.

An Orchestration Saga is a pattern where a central service, called an orchestrator, takes charge of guiding the entire transaction.

Think of the orchestrator as a project manager for your distributed transaction. It knows all the steps and coordinates them.

• It sends commands to participant services.
• It waits for responses (events) from those services.
• It decides the next step based on the responses.

Unlike a Choreography Saga where services react to events from each other, here the orchestrator is in control:

1. Orchestrator sends a command to Service A.
2. Service A performs its task and sends a completion event back to the orchestrator.
3. Orchestrator receives the event and sends a command to Service B.
4. This continues until the transaction is complete.

Let's use an online order as an example. An order might need to:

• Create an order in the Order Service.
• Reserve items in the Inventory Service.
• Process payment in the Payment Service.
• Initiate shipping in the Shipping Service.

An orchestrator would manage this sequence.

The orchestrator must keep track of the saga's current state and context. This means it needs to store information like:

• Which steps have completed?
• What data was returned by previous steps?
• What is the current status (e.g., PAYMENT_PENDING, INVENTORY_RESERVED)?

This state is crucial for deciding the next action or handling failures.

The orchestrator communicates using two main types of messages:

• Commands: Sent from the orchestrator to a service, telling it what to do (e.g., ReserveInventory).
• Events: Sent from a service back to the orchestrator, reporting the outcome (e.g., InventoryReserved, PaymentFailed).

This clear distinction helps maintain control.

What if a step fails? The orchestrator is responsible for compensation.

If the Payment Service fails, the orchestrator detects the PaymentFailed event. It then knows to send compensation commands to previous services, such as ReleaseInventory to the Inventory Service and CancelOrder to the Order Service.

Let's look at a simplified Java example. This isn't a full microservice, but it shows how an orchestrator's logic might flow, reacting to simulated service outcomes.

Notice how it tracks sagaState and triggers compensation if payment fails.

Orchestration Sagas offer clear benefits:

• Centralized Control: Easier to understand the overall flow.
• Simplified Participant Services: Services don't need to know about the entire saga; they just follow orchestrator commands.
• Easier Debugging: The orchestrator's state makes it simpler to trace issues.
• Explicit Compensation: Compensation logic is clearly defined within the orchestrator.

Which of the following statements accurately describe key characteristics of an Orchestration Saga?

You've learned about the Orchestration Saga pattern!

• It uses a central orchestrator service to manage a distributed transaction.
• The orchestrator sends commands to services and receives events as responses.
• It maintains the saga's state to know where it is in the transaction.
• It explicitly handles compensation by issuing rollback commands if a step fails.

This pattern provides clear control and simplifies participant services.