When planning an aiWARE instance, several important aspects of the configuration that are outlined below should be taken into consideration.
Before you begin
Instance design factors
| Configuration factors | Options |
| Internet access |
- If aiWARE will have access to the internet, you can create either a Veritone-managed instance that uses Veritone's cloud infrastructure, or a self-managed instance. Self-managed instances are for customers who are prepared to create, manage, and maintain their own cloud instances or on-premises server installations.
- No internet access means aiWARE must be installed and run on customer-owned hardware using the instructions for on-premises deployments.
|
| Hosting type |
- Veritone-managed instances are installed in the cloud and are managed by Veritone.
- Self-managed instances are typically used by customers who use either their own cloud environments, or who run network-isolated processing infrastructure.
|
| Infrastructure type when managed by Veritone |
- AWS EKS (Elastic Kubernetes Service) is the recommended type. EKS automates the management, scaling, and deployment of containerized applications.
- AWS (Amazon Web Services) allows customers to take advantage of Veritone's extensive AWS cloud provisioning knowledge.
|
| Instance type |
- The only currently supported option for instance type is AI Data + AI Processing, which includes cognitive processing and allows for data persistence, like long-term storage of job results, training of datasets, and engine benchmarking, among other tasks.
|
| Instance size |
- Small instances run on a single node.
- Medium instances run on four nodes.
- Large instances run on eleven nodes and are designed for resource-intensive, enterprise-scale deployments.
|
