Query-driven APIs.
WHAT IT IS
A query language where the client specifies exactly what data it wants.
WHAT PROBLEM IT ADDRESSES
Retrieving too much or too little from fixed endpoints, and making many requests to assemble a view.
WHAT IT PROVIDES
One endpoint A typed schema Clients requesting precisely the fields they need Introspection and generated documentation
WHAT IT COSTS
Caching becomes harder, since requests are not distinguished by address Complexity on the server Query cost that is difficult to predict
WHAT THE SEVERE PERFORMANCE PROBLEM IS
Resolving nested fields producing a query per item, repeatedly.
WHAT SOLVES IT
Batching and caching within a request.
WHAT TO IMPLEMENT ALWAYS
Query depth limits Complexity limits Timeouts
WHY
An unrestricted query can request deeply nested data and exhaust the server.
WHAT TO BE CAREFUL WITH
Introspection enabled publicly, revealing the entire schema Authorisation checked only at the top level
WHAT TO CHECK
Entitlement at every field resolving sensitive data.
WHEN TO USE IT
Many clients with differing data needs.
