BFF (Backend for Frontend) - The Savior of UX

Microservices are great for developers, but a nightmare for Clients (Web/Mobile) if not designed correctly. BFF is the "translator" layer that harmonizes these two worlds.

The Problem: The Client is Overwhelmed

Imagine the Home screen of an E-commerce app. To display it, the Mobile App needs to:

Call the User Service for personal info.
Call the Order Service for active deliveries.
Call the Recommendation Service for suggested products.
Call the Promotion Service for promotional banners.

Consequences: 5 simultaneous requests slow down page load, drain phone battery, and make error handling extremely difficult if one of them fails. Our team's Mobile Dev started complaining: "Why is doing microservices so miserable for me?".

1. What is a BFF?

BFF (Backend for Frontend) is an intermediary service, designed specifically for a particular type of Client (e.g., one for Web, one for Mobile). Its job is: Aggregating. Instead of calling 5 services, the Client now calls just 1 API to the BFF. The BFF will "collect" data from those 5 internal services, process it according to the specific format the UI needs, and return a single unified block of data.

2. Real-world Benefits

Network Optimization: Minimizes the number of requests from the Client. With unstable 4G/5G networks, this is the difference between "frozen" and "smooth."
Data Filtering: Internal microservices often return many redundant data fields. The BFF filters them, only sending back what the UI actually displays, helping to reduce payload size.
Hiding Changes: If we split the User Service into Profile Service and Account Service, the Mobile App doesn't need to care. Just update the logic in the BFF.

3. Trade-off: More Code Means More Work

For a 2-person team, maintaining an extra service (BFF) sounds irrational. Our practical formula: Instead of building a bulky Java/Go service as a BFF, we use the Server Side (API Routes) layer of Next.js or a thin Node.js Gateway. It allows Frontend Devs to write the data aggregation logic for themselves.

Lessons Learned

Don't Let the Client Know Too Much: The less the Client knows about the underlying microservice structure, the more flexible it is and the less it's affected by Backend refactors.
BFF Does Not Contain Business Logic: This is the most common mistake. The BFF should only be a "coordination" and "formatting" layer. Don't put financial calculations or heavy business logic here, otherwise, it will become a new Monolith.
Leverage GraphQL? If the team has the resources, GraphQL is a great tool for a BFF because it's born for querying data from multiple different sources. But if not, a simple REST API is enough.

Conclusion

BFF helps us keep the cleanliness of the Microservices underneath while still providing a smooth experience for the users on top. It is the necessary "buffer zone" so that distributed architecture doesn't become a burden for the user experience.

The Problem: The Client is Overwhelmed

Imagine the Home screen of an E-commerce app. To display it, the Mobile App needs to:

Call the User Service for personal info.

Call the Order Service for active deliveries.

Call the Recommendation Service for suggested products.

Call the Promotion Service for promotional banners.

1. What is a BFF?

2. Real-world Benefits

Network Optimization: Minimizes the number of requests from the Client. With unstable 4G/5G networks, this is the difference between "frozen" and "smooth."

Data Filtering: Internal microservices often return many redundant data fields. The BFF filters them, only sending back what the UI actually displays, helping to reduce payload size.

Hiding Changes: If we split the User Service into Profile Service and Account Service, the Mobile App doesn't need to care. Just update the logic in the BFF.

3. Trade-off: More Code Means More Work

Lessons Learned

Don't Let the Client Know Too Much: The less the Client knows about the underlying microservice structure, the more flexible it is and the less it's affected by Backend refactors.

BFF Does Not Contain Business Logic: This is the most common mistake. The BFF should only be a "coordination" and "formatting" layer. Don't put financial calculations or heavy business logic here, otherwise, it will become a new Monolith.

Leverage GraphQL? If the team has the resources, GraphQL is a great tool for a BFF because it's born for querying data from multiple different sources. But if not, a simple REST API is enough.