Self-Hosting Requirements
Infrastructure and prerequisites for successfully deploying Prisme.ai.
To successfully deploy and manage Prisme.ai within your infrastructure, you need to ensure that your environment meets specific prerequisites related to resources, services, and configurations.
Infrastructure Recommendations
Prisme.ai recommends the following baseline infrastructure to ensure smooth operation:
- Nodes: At least 3 Kubernetes nodes with 4 vCPU and 16GB RAM each【40:1†source】.
- Disk Space: Minimum 50GB recommended for persistent storage.
- High Availability: Use managed services or clusters with redundancy capabilities.
- Load Balancer: Ingress controller or cloud load balancer required for service routing.
Database Requirements
Ensure the following database setups for optimal performance and reliability:
-
MongoDB:
- Version: 6.x or newer
- Configuration: 3-node replica set, each with at least 2GB RAM, 2 vCPU, and 1,000 IOPS【32:2†source】.
-
Redis:
- Version: 5.x or newer
- Configuration: Separate instances recommended for cache and broker, each with at least 3GB RAM and 2 vCPU【32:2†source】.
-
Elasticsearch/OpenSearch:
- Version: 8.x or newer
- Configuration: 3-node cluster, each node with 16GB RAM and 4 vCPU【32:3†source】.
AI Products requirements
AI Products installation also require a vector Database which can be :
- Opensearch : Can be the same cluster as above, or a separated one
- Redis Stack : Redis 5.x+ with JSON and SEARCH modules, 3-node cluster with 5GB RAM and 2 vCPU each
Storage and Filesystems
Prisme.ai requires either a file or object storage for models & file uploads :
- RWX Filesystem: RWX volume with at least 50GB capacity, attached to a Kubernetes Persistent Volume Claims (PVC)
- S3-Compatible Storage
- Azure Blob Storage
- Google Cloud Storage
AI Products requirements
AI Products installation also require a RWX filesystem
Networking and DNS
DNS configuration requirements:
- Wildcard DNS: Essential for workspace routing (e.g.,
*.pages.yourdomain.com
). - Dedicated DNS: Required for API (
api.yourdomain.com
) and Studio (studio.yourdomain.com
)【40:2†source】.
In case wildcards DNS can’t be setup and AI Products need to be installed, the following DNS must be created :
- ai-knowledge.pages.yourdomain.com
- ai-knowledge-chat.pages.yourdomain.com
- ai-governance.pages.yourdomain.com
- teams-service.pages.yourdomain.com
- ai-collection.pages.yourdomain.com
- ai-insights.pages.yourdomain.com
However, other custom workspaces won’t be able to have their web pages reachable
Offline environments
AI Products requirements
AI Products installation in offline environments require :
- A npm registry reachable from the cluster
- If using internal / self-signed TLS certificates, they must be configured inside almost all microservices to avoid HTTPS errors
See AI Products requirements for more details
Authentication (SSO)
Prisme.ai supports multiple Single Sign-On (SSO) options:
- OIDC Providers: Compatible with Google, Azure, and other generic OIDC providers【35:1†source】.
- SAML Providers: Supported with HTTP-POST binding (e.g., Okta).
- Configuration: Managed via
authProviders.config.yml
for easy integration.
See SSO configuration documentation
Recommended Load Testing
Before going live, perform tailored load tests considering:
- Interactions per user: 4-10 per session.
- Simultaneous users: Support for 100 new users per second.
- API response time: Target 478 ms P95 for the first token using OpenAI API【32:1†source】.
Next Steps
Architecture
Explore detailed architectural diagrams and service relationships.
Cloud Providers
Learn about deploying Prisme.ai on AWS, Azure, GCP, and OpenShift.
Docker Deployment
Set up and configure Docker-based deployments.
Kubernetes Deployment
Understand Kubernetes deployments using Helm and operators.