Applying GraphQL to Solve Data Customization and Multi‑Request Issues in Qunar's Node BFF Service

In the Qunar domestic hotel service, the Node BFF layer faces three major problems: difficulty customizing data for different clients, manual maintenance of API documentation, and transmission of redundant fields, all of which hinder development efficiency and performance.

Case 1 – Data Customization Difficulty shows how the same endpoint returns all fields for PC and APP, while the mini‑program only needs a subset, leading to conditional code that removes fields based on the source and version:

hotelInfo: {
    name: "酒店",
    price: 232,
    imgUrl: "img.jpg",
    tags: [{ name: "亲子家庭" }],
    score: 4.2,
    rank: "XX酒店排名第X名"
}

if (source === "XIAOCHENGXU") {
    hotelInfo.tags = undefined;
    hotelInfo.score = undefined;
    hotelInfo.rank = undefined;
}

if (source === "XIAOCHENGXU") {
    hotelInfo.tags = undefined;
    if (version < 120) {
        hotelInfo.score = undefined;
    }
    hotelInfo.rank = undefined;
}

This approach quickly becomes unreadable and hard to maintain as requirements evolve.

Case 2 – Multi‑Request Splicing illustrates two ways to fetch hotel detail resources: issuing multiple RESTful requests for each resource or merging them into a single endpoint, which creates tight coupling and maintenance challenges.

// Method 1:
/api/detail {交通,点评}
/api/detailprice {房型,详情}
/api/clientShow {住客秀}
/api/recommendation {周边推荐}

// Method 2:
/api/detail {交通,点评}
/api/detailprice
if (source === "APP") {
    return {房型,详情,住客秀,周边推荐}
}
return {房型,详情}

Both methods have trade‑offs, but the underlying RESTful design limits flexibility.

GraphQL Introduction explains that GraphQL, created by Facebook in 2012, allows clients to request exactly the fields they need, automatically generates documentation, and enables free composition of resources through a single endpoint.

Migration Steps :

Define a GraphQL schema (SDL) for the hotel model:

type Hotel {
    name: String!
    imgUrl: String!
    price: Int!
    tags: [Tag]
    score: Float
    rank: String
}

type Tag {
    name: String!
}

Implement resolvers that fetch data from existing services or databases:

hotelResolver: (id: String) => {
    return db.getHotel(id);
}

Update client code to send GraphQL queries, e.g.:

query: hotel(id: "String") {
    name
    imgUrl
    price
    tags { name }
    score
    rank
}

By defining the schema, backend developers can focus on domain logic while frontend developers query precisely the data they need, eliminating over‑fetch and under‑fetch problems.

Practical Issues identified include increased backend design complexity, potential performance impacts, migration cost, and security concerns such as query depth limiting.

The article concludes that GraphQL is not a wholesale replacement for REST but a powerful complement for BFF layers that require frequent multi‑client data customization, offering a low‑risk, incremental migration path.