api-design-patterns（API 设计模式）

API Design Patterns

Design robust, scalable APIs using proven patterns for REST, GraphQL, and gRPC with proper versioning, authentication, and error handling.

Quick Reference

API Style Selection:

• REST: Resource-based CRUD, simple clients, HTTP-native caching
• GraphQL: Client-driven queries, complex data graphs, real-time subscriptions
• gRPC: High-performance RPC, microservices, strong typing, streaming

Critical Patterns:

• Versioning: URI (/v1/users), header (Accept: application/vnd.api+json;version=1), content negotiation
• Pagination: Offset (simple), cursor (stable), keyset (performant)
• Auth: OAuth2 (delegated), JWT (stateless), API keys (service-to-service)
• Rate limiting: Token bucket, fixed window, sliding window
• Idempotency: Idempotency keys, conditional requests, safe retry

See references/ for deep dives: rest-patterns.md, graphql-patterns.md, grpc-patterns.md, versioning-strategies.md, authentication.md

Core Principles

Universal API Design Standards

Apply these principles across all API styles:

1. Consistency Over Cleverness

• Follow established conventions for your API style
• Use predictable naming patterns (snake_case or camelCase, pick one)
• Maintain consistent error response formats
• Version breaking changes, never surprise clients

2. Design for Evolution

• Plan for versioning from day one
• Use optional fields with sensible defaults
• Deprecate gracefully with sunset dates
• Document breaking vs non-breaking changes

3. Security by Default

• Require authentication unless explicitly public
• Use HTTPS/TLS for all production endpoints
• Implement rate limiting and throttling
• Validate and sanitize all inputs
• Return minimal error details to clients

4. Developer Experience First

• Provide comprehensive documentation (OpenAPI, GraphQL schema)
• Return meaningful error messages with actionable guidance
• Use standard HTTP status codes correctly
• Include request IDs for debugging
• Offer SDKs and code generators

API Style Decision Tree

When to Choose REST

✅ Use REST when:

• Building CRUD-focused resource APIs
• Clients need HTTP caching (ETags, Cache-Control)
• Wide platform compatibility required (browsers, mobile, IoT)
• Simple, stateless client-server model fits
• Team familiar with HTTP/REST conventions

❌ Avoid REST when:

• Complex data fetching with nested relationships (N+1 queries)
• Real-time updates are primary use case
• Need strong typing and code generation
• High-performance RPC between microservices

Example Use Cases: Public APIs, mobile backends, traditional web services

When to Choose GraphQL

✅ Use GraphQL when:

• Clients need flexible, client-driven queries
• Complex data graphs with nested relationships
• Multiple client types with different data needs
• Real-time subscriptions required
• Strong typing and schema validation needed

❌ Avoid GraphQL when:

• Simple CRUD operations dominate
• HTTP caching is critical (GraphQL uses POST)
• File uploads are primary feature (requires extensions)
• Team lacks GraphQL expertise
• Performance optimization is complex (N+1 problem)

Example Use Cases: Client-facing APIs, dashboards, mobile apps with varied UIs

When to Choose gRPC

✅ Use gRPC when:

• Microservice-to-microservice communication
• High performance and low latency critical
• Bidirectional streaming needed
• Strong typing with Protocol Buffers
• Polyglot environments (language interop)

❌ Avoid gRPC when:

• Browser clients (limited support, needs grpc-web)
• HTTP/JSON required for compatibility
• Human-readable payloads preferred
• Simple request/response patterns

Example Use Cases: Internal microservices, streaming data, service mesh

REST API Patterns

Resource Naming

✅ Good: Plural nouns, hierarchical

GET    /users              # List users
GET    /users/123          # Get user
POST   /users              # Create user
PUT    /users/123          # Update user (full)
PATCH  /users/123          # Update user (partial)
DELETE /users/123          # Delete user
GET    /users/123/orders   # User's orders (sub-resource)

❌ Bad: Verbs, mixed conventions

GET    /getUsers           # Don't use verbs
POST   /user/create        # Don't use verbs
GET    /Users/123          # Don't capitalize
GET    /user/123           # Don't mix singular/plural

HTTP Status Codes

Success Codes:

• 200 OK: Successful GET, PUT, PATCH, DELETE with body
• 201 Created: Successful POST, return Location header
• 202 Accepted: Async operation started
• 204 No Content: Successful DELETE, no body

Client Error Codes:

• 400 Bad Request: Invalid input, validation error
• 401 Unauthorized: Missing or invalid authentication
• 403 Forbidden: Authenticated but insufficient permissions
• 404 Not Found: Resource doesn't exist
• 409 Conflict: State conflict (duplicate, version mismatch)
• 422 Unprocessable Entity: Semantic validation error
• 429 Too Many Requests: Rate limit exceeded

Server Error Codes:

• 500 Internal Server Error: Unexpected error
• 502 Bad Gateway: Upstream service error
• 503 Service Unavailable: Temporary outage
• 504 Gateway Timeout: Upstream timeout

Error Response Format

✅ Consistent error structure

{
  "error": {
    "code": "VALIDATION_ERROR",
    "message": "Invalid request parameters",
    "details": [
      {
        "field": "email",
        "message": "Invalid email format",
        "code": "INVALID_FORMAT"
      }
    ],
    "request_id": "req_abc123",
    "documentation_url": "https://api.example.com/docs/errors/validation"
  }
}

Pagination Patterns

Offset Pagination (simple, familiar):

GET /users?limit=20&offset=40

✅ Use for: Small datasets, admin interfaces ❌ Avoid for: Large datasets (skips become expensive), real-time data

Cursor Pagination (stable, efficient):

GET /users?limit=20&cursor=eyJpZCI6MTIzfQ
Response: { "data": [...], "next_cursor": "eyJpZCI6MTQzfQ" }

✅ Use for: Infinite scroll, real-time feeds, large datasets ❌ Avoid for: Random access, page numbers

Keyset Pagination (performant):

GET /users?limit=20&after_id=123

✅ Use for: Ordered data, database index friendly ❌ Avoid for: Complex sorting, multiple sort keys

See references/rest-patterns.md for filtering, sorting, field selection, HATEOAS

GraphQL Patterns

Schema Design

✅ Good: Clear types, nullable by default

type User {
  id: ID!                    # Non-null ID
  email: String!             # Required field
  name: String               # Optional (nullable by default)
  createdAt: DateTime!
  orders: [Order!]!          # Non-null array of non-null orders
}

type Query {
  user(id: ID!): User
  users(first: Int, after: String): UserConnection!
}

type Mutation {
  createUser(input: CreateUserInput!): CreateUserPayload!
}

input CreateUserInput {
  email: String!
  name: String
}

type CreateUserPayload {
  user: User
  userEdge: UserEdge
  errors: [UserError!]
}

Resolver Patterns

Avoid N+1 Queries with DataLoader:

import DataLoader from 'dataloader';

const userLoader = new DataLoader(async (userIds: string[]) => {
  const users = await db.users.findMany({ where: { id: { in: userIds } } });
  return userIds.map(id => users.find(u => u.id === id));
});

// Resolver batches queries automatically
const resolvers = {
  Order: {
    user: (order) => userLoader.load(order.userId)
  }
};

Query Complexity Analysis

Prevent expensive queries:

import { createComplexityLimitRule } from 'graphql-validation-complexity';

const server = new ApolloServer({
  schema,
  validationRules: [
    createComplexityLimitRule(1000, {
      onCost: (cost) => console.log('Query cost:', cost),
    }),
  ],
});

See references/graphql-patterns.md for subscriptions, relay cursor connections, error handling

gRPC Patterns

Service Definition

syntax = "proto3";

package users.v1;

service UserService {
  rpc GetUser (GetUserRequest) returns (User) {}
  rpc ListUsers (ListUsersRequest) returns (ListUsersResponse) {}
  rpc CreateUser (CreateUserRequest) returns (User) {}
  rpc StreamUsers (StreamUsersRequest) returns (stream User) {}
  rpc BidiChat (stream ChatMessage) returns (stream ChatMessage) {}
}

message User {
  string id = 1;
  string email = 2;
  string name = 3;
  google.protobuf.Timestamp created_at = 4;
}

message GetUserRequest {
  string id = 1;
}

message ListUsersRequest {
  int32 page_size = 1;
  string page_token = 2;
}

message ListUsersResponse {
  repeated User users = 1;
  string next_page_token = 2;
}

Error Handling

import (
    "google.golang.org/grpc/codes"
    "google.golang.org/grpc/status"
)

func (s *server) GetUser(ctx context.Context, req *pb.GetUserRequest) (*pb.User, error) {
    if req.Id == "" {
        return nil, status.Error(codes.InvalidArgument, "user ID is required")
    }

    user, err := s.db.GetUser(ctx, req.Id)
    if err != nil {
        if errors.Is(err, sql.ErrNoRows) {
            return nil, status.Error(codes.NotFound, "user not found")
        }
        return nil, status.Error(codes.Internal, "database error")
    }

    return user, nil
}

See references/grpc-patterns.md for streaming, interceptors, metadata, health checks

Versioning Strategies

URI Versioning (Simple, Explicit)

✅ Most common, easy to understand

GET /v1/users/123
GET /v2/users/123

Pros: Clear, easy to route, browser-friendly Cons: Couples version to URL, duplicates routes

Header Versioning (Clean URLs)

GET /users/123
Accept: application/vnd.myapi.v2+json

Pros: Clean URLs, version separate from resource Cons: Less visible, harder to test manually

Content Negotiation (Granular)

GET /users/123
Accept: application/vnd.myapi.user.v2+json

Pros: Resource-level versioning, backward compatible Cons: Complex, harder to implement

Version Deprecation Process

{
  "version": "1.0",
  "deprecated": true,
  "sunset_date": "2025-12-31",
  "migration_guide": "https://docs.api.com/v1-to-v2",
  "replacement_version": "2.0"
}

Include deprecation warnings:

HTTP/1.1 200 OK
Deprecation: true
Sunset: Sat, 31 Dec 2025 23:59:59 GMT
Link: <https://docs.api.com/v1-to-v2>; rel="deprecation"

See references/versioning-strategies.md for detailed migration patterns

Authentication & Authorization

OAuth 2.0 (Delegated Access)

Use for: Third-party access, user consent, token refresh

Authorization Code Flow (most secure for web/mobile):

1. Client redirects to /authorize
2. User authenticates, grants permissions
3. Auth server redirects to callback with code
4. Client exchanges code for access token
5. Client uses access token for API requests

# Request token
POST /oauth/token
Content-Type: application/x-www-form-urlencoded

grant_type=authorization_code
&code=AUTH_CODE
&redirect_uri=https://client.com/callback
&client_id=CLIENT_ID
&client_secret=CLIENT_SECRET

# Response
{
  "access_token": "eyJhbGc...",
  "token_type": "Bearer",
  "expires_in": 3600,
  "refresh_token": "tGzv3JOkF0XG5Qx2TlKWIA",
  "scope": "read write"
}

# Use token
GET /v1/users/me
Authorization: Bearer eyJhbGc...

JWT (Stateless Auth)

Use for: Microservices, stateless API auth, short-lived tokens

✅ Good: Minimal claims, short expiry

{
  "sub": "user_123",
  "iat": 1516239022,
  "exp": 1516242622,
  "scope": "read:users write:orders"
}

Validation:

import jwt from 'jsonwebtoken';

const token = req.headers.authorization?.split(' ')[1];
const payload = jwt.verify(token, process.env.JWT_SECRET);
req.userId = payload.sub;

API Keys (Service-to-Service)

Use for: Server-to-server, CLI tools, webhooks

GET /v1/users
X-API-Key: sk_live_abc123...

# Or query parameter (less secure)
GET /v1/users?api_key=sk_live_abc123

Key Practices:

• Prefix keys with environment (sk_live_, sk_test_)
• Hash keys before storage (bcrypt, scrypt)
• Allow key rotation without downtime
• Support multiple keys per user
• Rate limit per key

See references/authentication.md for API key rotation, scopes, RBAC

Rate Limiting

Token Bucket (Burst-Friendly)

Bucket: 100 tokens, refill 10/second
Request costs 1 token
Allows bursts up to bucket size

Headers:

HTTP/1.1 200 OK
X-RateLimit-Limit: 100
X-RateLimit-Remaining: 73
X-RateLimit-Reset: 1640995200

429 Response:

HTTP/1.1 429 Too Many Requests
Retry-After: 60
X-RateLimit-Limit: 100
X-RateLimit-Remaining: 0
X-RateLimit-Reset: 1640995200

{
  "error": {
    "code": "RATE_LIMIT_EXCEEDED",
    "message": "Rate limit exceeded. Try again in 60 seconds.",
    "limit": 100,
    "reset_at": "2025-01-01T00:00:00Z"
  }
}

Sliding Window (Fair Distribution)

Counts requests in rolling time window. More accurate than fixed window.

Per-User vs Per-IP

• Per-User: Authenticated requests, fair quotas
• Per-IP: Unauthenticated requests, prevent abuse
• Combined: Both limits, take stricter

Idempotency

Idempotent Methods (HTTP Spec)

Naturally Idempotent: GET, PUT, DELETE, HEAD, OPTIONS Not Idempotent: POST, PATCH

Idempotency Keys

Make POST requests idempotent:

POST /v1/payments
Idempotency-Key: uuid-or-client-generated-key
Content-Type: application/json

{
  "amount": 1000,
  "currency": "USD",
  "customer": "cust_123"
}

Server behavior:

1. First request: Process and store result with key
2. Duplicate request (same key): Return stored result (200 or 201)
3. Different request (same key): Return 409 Conflict

Implementation:

const idempotencyKey = req.headers['idempotency-key'];
if (idempotencyKey) {
  const cached = await redis.get(`idempotency:${idempotencyKey}`);
  if (cached) {
    return res.status(cached.status).json(cached.body);
  }
}

const result = await processPayment(req.body);
await redis.setex(`idempotency:${idempotencyKey}`, 86400, {
  status: 201,
  body: result
});

Conditional Requests

Use ETags for safe updates:

# Get resource with ETag
GET /v1/users/123
Response: ETag: "abc123"

# Update only if unchanged
PUT /v1/users/123
If-Match: "abc123"

# 412 Precondition Failed if ETag changed

Caching Strategies

HTTP Caching Headers

# Public, cacheable for 1 hour
Cache-Control: public, max-age=3600

# Private (user-specific), revalidate
Cache-Control: private, must-revalidate, max-age=0

# No caching
Cache-Control: no-store, no-cache, must-revalidate

ETag Validation

# Server returns ETag
GET /v1/users/123
Response:
  ETag: "33a64df551425fcc55e4d42a148795d9f25f89d4"
  Cache-Control: max-age=3600

# Client conditional request
GET /v1/users/123
If-None-Match: "33a64df551425fcc55e4d42a148795d9f25f89d4"

# 304 Not Modified if unchanged (saves bandwidth)
HTTP/1.1 304 Not Modified

Last-Modified

GET /v1/users/123
Response:
  Last-Modified: Wed, 21 Oct 2025 07:28:00 GMT

# Conditional request
GET /v1/users/123
If-Modified-Since: Wed, 21 Oct 2025 07:28:00 GMT

# 304 Not Modified if not modified

Webhooks

Event Delivery

POST https://client.com/webhooks/payments
Content-Type: application/json
X-Webhook-Signature: sha256=abc123...
X-Webhook-Id: evt_abc123
X-Webhook-Timestamp: 1640995200

{
  "id": "evt_abc123",
  "type": "payment.succeeded",
  "created": 1640995200,
  "data": {
    "object": {
      "id": "pay_123",
      "amount": 1000,
      "status": "succeeded"
    }
  }
}

Signature Verification

import crypto from 'crypto';

function verifyWebhookSignature(
  payload: string,
  signature: string,
  secret: string
): boolean {
  const expectedSignature = crypto
    .createHmac('sha256', secret)
    .update(payload)
    .digest('hex');
  return crypto.timingSafeEqual(
    Buffer.from(signature),
    Buffer.from(`sha256=${expectedSignature}`)
  );
}

Retry Strategy

• Exponential backoff: 1s, 2s, 4s, 8s, 16s, 32s, 64s
• Timeout: 5-30 seconds per attempt
• Max attempts: 3-7 attempts
• Dead letter queue: Store failed events
• Manual retry: UI for re-sending failed events

API Documentation

OpenAPI/Swagger (REST)

openapi: 3.0.0
info:
  title: User API
  version: 1.0.0
paths:
  /users/{id}:
    get:
      summary: Get user by ID
      parameters:
        - name: id
          in: path
          required: true
          schema:
            type: string
      responses:
        '200':
          description: Successful response
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/User'
        '404':
          description: User not found
components:
  schemas:
    User:
      type: object
      required: [id, email]
      properties:
        id:
          type: string
        email:
          type: string
          format: email
        name:
          type: string

GraphQL Schema (Self-Documenting)

GraphQL introspection provides automatic documentation. Use descriptions:

"""
Represents a user account in the system.
Created via the createUser mutation.
"""
type User {
  """Unique identifier for the user"""
  id: ID!

  """Email address, must be unique"""
  email: String!

  """Optional display name"""
  name: String
}

API Documentation Best Practices

1. Interactive examples: Provide working code samples
2. Authentication guide: Step-by-step auth setup
3. Error catalog: Document all error codes with examples
4. Rate limits: Clearly state limits and headers
5. Changelog: Track breaking and non-breaking changes
6. Migration guides: Version upgrade instructions
7. SDKs: Provide client libraries for popular languages

Anti-Patterns

❌ Over-fetching (REST): Returning entire objects when fields are unused ✅ Solution: Support field selection (?fields=id,name,email)

❌ Under-fetching (REST): Requiring multiple requests for related data ✅ Solution: Support expansion (?expand=orders,profile) or use GraphQL

❌ Chatty APIs: Too many round-trips for common operations ✅ Solution: Batch endpoints, compound documents, or GraphQL

❌ Ignoring HTTP semantics: Using GET for mutations, wrong status codes ✅ Solution: Follow HTTP spec, use correct methods and status codes

❌ Exposing internal structure: URLs/schemas mirror database ✅ Solution: Design resource-oriented APIs independent of storage

❌ Missing versioning: Breaking changes without version increments ✅ Solution: Version from day one, never break existing versions

❌ Poor error messages: Generic "An error occurred" ✅ Solution: Specific, actionable error messages with codes

❌ No rate limiting: APIs vulnerable to abuse ✅ Solution: Implement rate limiting from the start

Testing Strategies

Contract Testing

// Pact contract test
import { PactV3 } from '@pact-foundation/pact';

const provider = new PactV3({
  consumer: 'FrontendApp',
  provider: 'UserAPI'
});

it('gets a user by ID', () => {
  provider
    .given('user 123 exists')
    .uponReceiving('a request for user 123')
    .withRequest({
      method: 'GET',
      path: '/users/123'
    })
    .willRespondWith({
      status: 200,
      body: { id: '123', email: 'user@example.com' }
    });
});

Load Testing

// k6 load test
import http from 'k6/http';
import { check } from 'k6';

export const options = {
  stages: [
    { duration: '30s', target: 20 },
    { duration: '1m', target: 20 },
    { duration: '10s', target: 0 }
  ],
  thresholds: {
    http_req_duration: ['p(95)<500'], // 95% under 500ms
    http_req_failed: ['rate<0.01']    // <1% errors
  }
};

export default function () {
  const res = http.get('https://api.example.com/users');
  check(res, {
    'status is 200': (r) => r.status === 200,
    'response time < 500ms': (r) => r.timings.duration < 500
  });
}

Related Skills

• graphql: Deep GraphQL schema design, resolvers, Apollo Server
• typescript: Type-safe API clients and servers
• nodejs-backend: Express/Fastify REST API implementation
• django: Django REST Framework patterns
• fastapi: FastAPI Python REST/GraphQL APIs
• flask: Flask-RESTful patterns

References

• rest-patterns.md: Deep REST coverage (HATEOAS, filtering, field selection)
• graphql-patterns.md: GraphQL subscriptions, relay cursor connections, federation
• grpc-patterns.md: Streaming patterns, interceptors, service mesh integration
• versioning-strategies.md: Detailed versioning approaches and migration patterns
• authentication.md: OAuth flows, JWT best practices, API key rotation, RBAC

Additional Resources

• REST API Design Rulebook - O'Reilly REST guide
• GraphQL Best Practices - Official GraphQL guide
• gRPC Best Practices - Official gRPC guide
• RFC 7807: Problem Details for HTTP APIs - Standard error format
• OpenAPI Specification - REST documentation standard