Saga — what do you need it for, what do you operate with?

Saga is a pattern for managing distributed transactions that ensures data consistency in a microservices architecture without using global locks.

The main idea: split a large transaction into a series of local transactions, each with a compensating operation in case of rollback.

I have used Saga to coordinate processes, for example, when placing an order, where multiple services need to be updated (inventory, payments, delivery). If one operation fails, a compensating transaction is triggered to undo the previous steps.

What I operate with:

• Saga orchestration (a centralized controller that manages steps and compensating transactions)
• Saga choreography (each service reacts to events and initiates its own steps)

In Go, I often implement Saga using events and message queues (e.g., Kafka, NATS), and also use libraries for managing transaction states and timeouts.