Skip to main content

Requirements

There are some requirements that need to be met before you can deploy the application.

  • Fully qualified domain name
  • SSL certificate
  • Managed cloud (AWS/Azure/GC)
  • Managed Kubernetes Cluster (EKS/AKS/GKE)
info

If you are using Penfield Azure Marketplace offering to deploy product then Cloud infrastructure gets auto created, but you still need to fulfil other requirements.

Product Requirements:​

Fully qualified domain name (FQDN)​

  • Product will be deployed on a fully qualified domain name (FQDN) (example: penfieldai.example.com)

SSL Certificate​

  • SSL certificate for the FQDN is required. Depend on the Cloud Provider you use, you can either use free cloud provided SSL certificate (like AWS ACM) or you can use your own SSL certificate.
info

AWS provide free SSL that you can use with AWS services.

Infra Requirements:​

Cloud requirements​

  • AWS
  • Azure

Network requirements​

  • Network CIDR: /24 (minimum), /16 (recommended)
  • Subnets: 3 (public, private-web, private-database), each with 2 Availability zone (public and private-database subnets is optional, if you not have any public LoadBalancer endpoints and choose to deploy DB inside K8 cluster.)

Kubernetes cluster requirements​

  • K8 cluster version: 1.28 or newer
  • Node size: 8 CPU, 32 GB memory, 50GB SSD [AWS: m5.2xlarge, Azure: Standard_D8_v4]
  • Number of Nodes: Min:2, Max: Varies (with Auto Scaling) [For Dev/POC Max: 3, for PROD Max: 5]
  • Storage: 1-2 TB (on demand as required) [will be auto created inside k8 cluster using persistence volumes]

PostgreSQL requirements​

PSQL is required for penfield app to deploy, but you can choose to have a dedicated DB server or deploy inside K8 cluster. If you choose to deploy in K8 cluster then this is not needed.

  • Instance size: 2 CPU, 8 GB memory, 200GB SSD (minimum), 4 CPU, 32 GB memory, 200GB SSD (recommended with expandable storage)
  • Availability zone: 2 (minimum)
  • Replica: Yes

Azure OpenAI​

Azure OpenAI service is needed. You need to create GPT model in azure OpenAI instance.

Next steps:​

  • Secure fully qualified domain name: Secure FQDN example: penfieldai.example.com, To point A record for the domain you can get the IP once LoadBalancer is created.
  • Get the certificates related to fully qualified domain: For the above FQDN, you need to get the cert either from ACM or generate your own, you will need those certs when you have to setup infrastructure components.
  • Set up cloud infrastructure (Virtual network, K8 cluster, postgreSQL database).
  • Set up Azure OpenAI (Virtual network, K8 cluster, postgreSQL database).
  • Install Kubectl, Helm cli and helmfile: Download and install these cli tool on your local. Follow instructions at Install tools.
  • Set up addons on Kubernetes cluster.
  • Get credentials to pull docker images and helm charts, Contact Penfield.AI to receive the credentials.
  • Set up application on Kubernetes cluster. eg: nginx controller, kafka, postgres, kratos, airflow, milvus, minio, redis, Penfield-app.