This content is currently WIP. Diagrams, content, and structure are subject to change.
Kubernetes serves as the robust foundation of the C3 Agentic AI Platform’s infrastructure, providing powerful capabilities for deploying, scaling, and managing containerized services. This enterprise-grade orchestration system enables the platform to deliver consistent performance, reliability, and security across diverse deployment environments.
The C3 Agentic AI Platform runs on Kubernetes, an industry-standard container orchestration system that automates the deployment, scaling, and management of containerized applications. This architecture provides several key benefits:The diagram above shows how the C3 AI Infrastructure Layer uses Kubernetes to manage resources. This containerized approach enables:
Consistent deployment across cloud providers and on-premises environments
Automatic scaling based on workload demands
Self-healing capabilities that recover from failures
Resource optimization through efficient allocation of compute resources
The platform can be deployed on major cloud providers including AWS, Azure, and Google Cloud. Each deployment leverages cloud-native services while maintaining a consistent application experience:
AWS deployments use EKS (Elastic Kubernetes Service) with integration to services like S3, RDS, and CloudWatch
Azure deployments use AKS (Azure Kubernetes Service) with integration to Azure Blob Storage, Azure SQL, and Azure Monitor
Google Cloud deployments use GKE (Google Kubernetes Engine) with integration to Cloud Storage, Cloud SQL, and Cloud Monitoring
The C3 Agentic AI Platform scales automatically to handle varying workloads:
Horizontal Pod Autoscaling (HPA) adds or removes container instances based on CPU, memory, or custom metrics
Vertical Pod Autoscaling (VPA) adjusts resource requests and limits based on actual usage
Cluster Autoscaling adds or removes nodes to accommodate changing resource demands
For example, when processing data from thousands of wind turbines, the platform can automatically scale up data ingestion services to handle the increased load, then scale them back down when the processing is complete.