Skip to main content

Level 3

Here we dive deeper into the most complex and critical components from Level 2, showing their internal implementation details and patterns.

Saga Orchestrator - Internal Implementation​

The Saga Orchestrator is a critical component that manages distributed transactions across multiple services. It implements the Saga pattern to ensure data consistency in a microservices environment.

Saga Definition​

Purpose/Responsibility Defines the workflow steps, compensation actions, and business rules for distributed transactions.

Implementation Details

  • Fluent API for saga definition
  • Step dependencies and ordering
  • Compensation action mapping
  • Timeout configuration

Code Structure

src/Services/Ordering/
├── Sagas/
│   ├── OrderProcessingSaga.cs
│   ├── PaymentProcessingSaga.cs
│   └── InventoryReservationSaga.cs
├── SagaDefinitions/
│   ├── OrderSagaDefinition.cs
│   └── PaymentSagaDefinition.cs

State Manager​

Purpose/Responsibility Tracks saga execution progress, manages state transitions, and ensures consistency.

Implementation Details

  • State machine implementation
  • Event sourcing for state changes
  • Optimistic concurrency control
  • State snapshot capabilities

Transaction Coordinator​

Purpose/Responsibility Orchestrates the execution of saga steps across multiple services.

Implementation Details

  • Command dispatching
  • Response correlation
  • Parallel step execution
  • Dependency resolution

Compensation Handler​

Purpose/Responsibility Executes compensation actions when saga steps fail, ensuring system consistency.

Implementation Details

  • Reverse operation execution
  • Compensation ordering
  • Partial failure handling
  • Idempotent compensation

Search Engine - Vector Search Implementation​

The Search Engine implements sophisticated AI-powered search using vector embeddings and semantic similarity.

Document Indexer​

Purpose/Responsibility Processes product documents and creates searchable index entries with vector embeddings.

Implementation Details

  • Document parsing and normalization
  • Metadata extraction
  • Batch processing capabilities
  • Index update strategies

Processing Pipeline

  1. Document preprocessing
  2. Text extraction and cleaning
  3. Embedding generation
  4. Vector storage
  5. Index optimization

Embedding Generator​

Purpose/Responsibility Converts text content into high-dimensional vector representations using machine learning models.

Implementation Details

  • Integration with OpenAI/Azure OpenAI
  • Batch embedding generation
  • Caching of embeddings
  • Model version management

Quality Characteristics

  • Sub-second embedding generation
  • High-dimensional vector space (1536 dimensions)
  • Semantic similarity preservation
  • Model consistency across updates

Query Processor​

Purpose/Responsibility Processes user search queries and converts them into vector search operations.

Implementation Details

  • Query parsing and normalization
  • Intent recognition
  • Filter application
  • Hybrid search (vector + keyword)

Search Features

  • Semantic similarity search
  • Faceted search with filters
  • Auto-completion suggestions
  • Search result personalization