The Requirements: Scaling a knowledge graph up to 20TB in size
Knowledge graph support: Altana needed the capability to model all the diverse entities in the supply chain, the linkages between them and different sources, and be able to navigate and traverse the graph in a consistent manner. Ian knew a knowledge graph approach was the best way to solve this.
Scalability: Ian estimated that the graph they were looking to build, Altana Atlas, could grow up to 20 TB in size over the course of a year. Making this even more challenging is the fact that every Altana customer required its own dedicated copy of Atlas to protect customer privacy.
Performance: Altana needed a way query with very low latency, even at scale. This included the ability to join terabytes of data across multiple graph hops to perform complex multi-tier analyses.
Integration with ML pipeline: Altana’s machine learning pipeline requires data to flow from Databricks into a graph database, using an Apache Spark connector.
High availability: Given the mission-critical nature of supply chains, Altana needed to ensure that its customers’ supply chain data is continuously available via multiple replicas.
Compatibility with AWS, Azure, and VMware: Altana needed its solution to run on multiple cloud providers, including AWS and Azure, as well as self-managed infrastructure provided by VMware. This would allow Altana to run its service close to its customers’ other supply chain applications, reducing expensive data egress charges resulting from sending data from one cloud provider to another.
Cloud-based: Altana needed a managed database so it could focus on innovation rather than rote database maintenance tasks.
Security: Due to the sensitive nature of its customers’ data, Altana needed encryption at rest and in transit to meet data rights and privacy regulations and standards.
The Implementation: A global platform to build trusted networks
Using ArangoGraph as the foundation, Altana built the Altana Atlas, a federated platform powered by an AI model that connects and learns from the world’s public and non-public data to provide a dynamic, intelligent model of the supply chain while ensuring the sovereignty, security, and privacy of the underlying data. It allows:
- Enterprises to build trusted multi-tier supply chain networks to power planning, procurement, risk management, ESG, and compliance systems from a living map of the supply chain;
- Governments to ‘see’ across borders and secure their supply chains by connecting the dots between the global commercial network and threats to national interests and corporate compliance;
- Logistics providers and customs authorities to revolutionize the cross-border movement of goods by digitalizing and automating cross-border compliance.
This platform is built on the following data architecture:
