Deploy LLM Gateway Only

Architecture

The architecture of Truefoundry Gateway comprises the truefoundry helm chart, which comprises of the following components:

  1. Frontend
  2. Backend Microservice
  3. Postgres DB to store all metadata - this is created on Kubernetes in dev mode. However, for production setups, we recommend using a managed database like AWS RDS or GCP Cloud SQL.
  4. Queue to handle request logging and configuration propagation.
  5. Clickhouse DB to store all request logs.

The helm chart can be completely deployed within your own VPC. This will talk to the Truefoundry authentication/licensing server in Truefoundry VPC to validate the license key and users. The only information that is communicated to the central licensing server is the user's email address who logs into the platform and analytics on the number of requests made through the gateway.

Prerequisites for Installation

  1. Egress access to auth.truefoundry.com and analytics.truefoundry.com
  2. Domain to map the ingress of the frontend and gateway
  3. Tenant, Licence key, and image pull secret from TrueFoundry team

Installation Instructions

  1. Create a values file as given below and replace relevant values:
    global:
      # Ask TrueFoundry team to provide these
      tenantName: <TENANT_NAME>
      tfyApiKey: <LICENCE_KEY>
      truefoundryImagePullConfigJSON: <TRUEFOUNDRY_IMAGE_PULL_SECRET>
      
      # Domain to map the platform to
      controlPlaneURL: https://example.com
      resourceTier: small
    
    truefoundryFrontendApp:
      # Choose the service mesh: we support both istio and k8s ingress, enable as per your setting
      # Replace `example.com` with your Domain
      istio:
        virtualservice:
          hosts:
            - example.com
          enabled: false
          gateways:
            - istio-system/tfy-wildcard
      ingress:
        hosts:
          - example.com
        enabled: false
        annotations: {}
        ingressClassName: nginx
    tags:
      llmGateway: true
      llmGatewayRequestLogging: true
    # Enable devMode
    devMode:
      enabled: true
    # Disable few depenencies for only LLM Gateway setup
    tfyBuild:
      enabled: false
    tfyController:
      enabled: false
    mlfoundryServer:
      enabled: false
    tfy-buildkitd-service:
      enabled: false
  2. Add helm repo
    helm repo add truefoundry https://truefoundry.github.io/infra-charts
  3. Install the helm chart with your values file
    helm upgrade --install truefoundry truefoundry/truefoundry -n truefoundry --create-namespace -f truefoundry-values.yaml