Skip to main content

Architecture

This document explains Bundleport's architecture and how requests flow through the system.

High-Level Architecture​

Request Flow​

1. Search Request​

2. Booking Request​

Key Components​

API Gateway​

Responsibilities:

  • Authentication (API key validation and automatic token transformation)
  • Rate limiting (per service account, per endpoint)
  • Request routing
  • Response formatting
  • Logging and monitoring

The API Gateway acts as the entry point for all API requests, handling security, rate limiting, and routing before requests reach backend services.

Authentication (product view):

  1. Your client sends Authorization: ApiKey <your-key> on each request.
  2. The gateway validates the key, applies rate limits, and routes to the right service.
  3. Backend services enforce scopes and business rules.

You do not need to obtain or refresh Bearer tokens in your application—the gateway handles token exchange internally. For implementation details (caching, JWT), see Authentication.

Connect Hotels Service​

Responsibilities:

  • Request validation
  • Business rule application
  • Data normalization
  • Provider orchestration
  • Response aggregation
  • Deduplication and ranking

Key Features:

  • Multi-provider aggregation
  • Parallel request execution
  • Fault tolerance (partial failures)
  • Caching strategies

Provider Integrations​

Responsibilities:

  • Protocol translation (REST, SOAP, etc.)
  • Data mapping (provider format → Bundleport format)
  • Error handling
  • Retry logic
  • Connection pooling

Supported Protocols:

  • REST APIs
  • SOAP/XML
  • Binary protocols
  • Custom integrations

Content Service​

Responsibilities:

  • Hotel catalog management
  • Content synchronization
  • Data quality assurance
  • Multi-language support
  • Media management

Data Flow​

Normalization Process​

Example:

  • Provider A: hotel_id, room_type, meal_plan
  • Provider B: propertyCode, accommodationType, boardBasis
  • Normalized: hotel.code, room.code, boardCode

Deduplication​

When the same hotel appears from multiple providers:

  1. Identify - Match hotels by location, name, or provider mapping
  2. Consolidate - Merge options from all providers
  3. Rank - Apply business rules to select best option
  4. Present - Return single result with provider metadata

Multi-Provider Aggregation​

Parallel Execution​

All providers are queried in parallel for faster responses:

Wall-clock time is driven by the slowest successful supplier (not the sum of all calls).

Partial Failures​

If one provider fails, others continue:

Caching Strategy​

Content API​

  • Cache Duration: Hours to days
  • Cache Key: Connection code + content type
  • Invalidation: On content updates

Booking API​

  • Cache Duration: None (always real-time)
  • Reason: Availability and pricing change frequently

Scalability​

Horizontal Scaling​

  • API Gateway: Stateless, scales horizontally
  • Services: Stateless, scales horizontally
  • Database: Read replicas for content queries

Performance Optimizations​

  • Connection pooling to providers
  • Parallel request execution
  • Response caching (where appropriate)
  • Database query optimization

Security​

Authentication​

  • API key validation at gateway
  • Automatic ApiKey → Bearer token transformation (transparent to clients)
  • Service account scopes
  • IP allowlisting (optional)

Data Protection​

  • HTTPS/TLS for all communications
  • PCI compliance for payment data
  • Secure credential storage

Monitoring & Observability​

Metrics​

  • Request rates
  • Response times
  • Error rates
  • Provider performance

Tracing​

  • Request IDs for end-to-end tracking
  • Per-provider response times
  • Error tracking

Logging​

  • Request/response logging
  • Error logging
  • Audit trails

Next Steps​