Is Data Consistency Guaranteed Across Services?

Is data consistency between services guaranteed?

Background/Overview

In microservices or loosely coupled systems, ensuring consistency becomes difficult, so it is necessary to clarify "when consistency should be achieved" during the design phase.

---

Examples

• Adopt a consistency strategy based on event-driven architecture + eventual consistency.
• Instead of synchronizing a status update bidirectionally between the Task service and the Document service, ensure consistency through a unidirectional data flow, such as having only the Task service perform the update and the Document service only subscribe to it.
• Prevent inconsistencies by sending events after updates and validating them on the receiving end.

---

Common Failure Examples

• Data becomes inconsistent due to bidirectional updates, making it unclear which information is the latest.
• Processing order is not guaranteed, leading to data being overwritten by older events.
• Read discrepancies occur due to replica lag.

---

FAQ

Q. Is eventual consistency really okay?

A. It depends on the case, but for many business operations, "consistency within a certain timeframe" is often sufficient rather than immediate consistency. It is important to consider this, define it, and document it clearly.

Q. Should bidirectional synchronization be avoided?

A. As a general rule, yes. You should avoid it because it leads to "synchronization hell" unless you adopt a unidirectional update flow.

---

Related Perspectives

• Is the necessity of distributed transactions for ensuring consistency well-organized?
• Is the design built to consider the impact of asynchronous event delays and retry processing?
• Has the impact of failures in dependent external services been considered?
• Is the scope of application for asynchronous and synchronous processing appropriate?

---

📘 This article is part of the SaaS Design Review Perspective Checklist (2025 Edition).

0 Design perspective category: 🔁 Events / Asynchronous Processing
0 Level: Structure (Structure, Principles, Responsibility Organization Level)
0 Recommended readers: Design beginners / Teams in the design phase / Reviewers

🧭 Click here to read other perspectives as well.