Mapping a GraphQL API to Your Data Sources
Select Sources
The module sources vary depending on the implementation of your GraphQL API. APIkit for GraphQL provides two kind of sources:
-
Data Fetchers
-
Data Loaders
Your selection depends on your implementation needs.
GraphQL is transport-layer agnostic, so you choose the main connector listener depending on the protocol that you want to serve your data.
Add a Listener
The inbound endpoint that triggers the main application flow is the connector listener specific to the client-server protocol of your choice. GraphQL often communicates over HTTP, so APIkit scaffolder generates an HTTP listener by default, but you can choose another source for serving your data over a different protocol.
Add a Data Fetcher
Data fetchers are sources of Mule flows. They fulfill the data of a single-object field in a query result. The combination of the following key parts identify them:
-
Object type name: One of the object types in your schema
-
Field name: One of the fields that are part of the object type
You can add data fetchers for every object field defined on the schema, regardless of the field type. If the Mule application does not have a data fetcher configured for a field, the parent type populates the field value. Otherwise, the client-requested data is not fulfilled, resulting in null.
Add a Data Loader
As data fetchers, data loaders are also sources of Mule flows. They batch requests related to an object type. Unlike data fetchers, a unique key identifies them:
-
Object type name: One of the object types in your schema
Data loaders solve the optimization issue of the N+1 requests when implementing GraphQL APIs. This issue occurs when N additional query statements are executed to fetch data that could have been retrieved when performing the first query.
At run time, the module looks for a data fetcher first. If the data fetcher is not available, the module looks for a data loader. Specifically, in case a data fetcher and a data loader exist for the same object type, the module uses the data fetcher as the first choice for resolving a field. This process can impact your API performance because the data fetcher keeps getting invoked N+1 times.
Choose Operations
The module provides a set of operations that are fundamental for the correct design and execution of a GraphQL API implementation:
-
Route
-
Serialize
-
Serialize Batch
Route a Query
The Route operation is the entry point of a GraphQL service. It parses, validates, and executes the query by traversing the graph and calling the data fetcher and data loaders required to fulfill the query result.