# Architecture overview
<a name="data-mesh-architecture-overview-section"></a>

The Automotive Data Mesh follows a domain-oriented decentralized architecture with centralized governance.

![Automotive Data Mesh Architecture](http://docs.aws.amazon.com/guidance/latest/automotive-data-platform-on-aws/images/mesh.png)


## High-Level Architecture
<a name="high-level-architecture"></a>

The architecture consists of:

1.  **Foundation Layer**: VPC networking with private subnets and VPC endpoints

1.  **Catalog Layer**: DataZone domain for data asset management

1.  **Compute Layer**: Distributed processing across Glue, SageMaker, Athena, and EMR

1.  **Governance Layer**: Centralized policies with federated enforcement

1.  **Collaboration Layer**: SageMaker Unified Studio for team workspaces

## Multi-Source Data Integration
<a name="multi-source-data-integration"></a>

The platform integrates data from diverse automotive sources:
+  **Vehicle Telemetry**: IoT Core for real-time data, S3 for historical data
+  **Customer Data**: CRM systems, service records, interaction history
+  **Sales and Inventory**: Dealer management systems, supply chain data
+  **External Data**: Weather, traffic, maps, and third-party enrichment
+  **Operational Data**: Manufacturing, logistics, warranty claims

## Networking Architecture
<a name="networking-architecture"></a>

Secure, private networking foundation:
+  **VPC with 3 Availability Zones**: High availability and fault tolerance
+  **Private Subnets**: All compute resources in private subnets
+  **NAT Gateway**: Controlled outbound internet access
+  **VPC Endpoints**: Private connectivity to AWS services (S3, Glue, SageMaker, Athena, Redshift, etc.)
+  **Security Groups**: Fine-grained network access control
+  **Network Isolation**: Separation between data domains