GenAIInfra is the containerization and cloud native suite for OPEA, including artifacts to deploy GenAIExamples in a cloud native way, which can be used by enterprise users to deploy to their own cloud.
The GenAIInfra repository is organized under four main directories, which include artifacts for OPEA deploying:
| Directory | Purpose |
|---|---|
authN-authZ |
Authentication and Authorization scenarios for OPEA. |
helm-charts |
Helm charts for deploying GenAIComponents on Kubernetes. |
microservices-connector |
GenAI Microservices Connector (GMC) supports the launching, monitoring, and updating of GenAI microservice chains, such as those in GenAIExamples on Kubernetes. It essentially supports a Kubernetes Custom Resource Definition for GenAI chains/pipelines that may be comprised of sequential, conditional, and parallel steps. |
kubernetes-addons |
Deploy Kubernetes add-ons for OPEA. |
proxy |
OPEA Pipeline Proxy is an enhancement of the default Istio proxy with additional features designed specifically for OPEA RAG pipelines. |
scripts |
Scripts for testing, tools to facilitate OPEA deployment, and etc. |
GenAIInfra uses Kubernetes as the cloud native infrastructure. Follow the steps below to prepare the Kubernetes environment.
Follow Kubernetes official setup guide to setup Kubernetes. We recommend to use Kubernetes with version >= 1.27.
There are different methods to setup Kubernetes production cluster, such as kubeadm, kubespray, and more.
NOTE: We recommend to use containerd when choosing the container runtime during Kubernetes setup. Docker engine is also verified on Ubuntu 22.04 and above.
(Optional) To run GenAIInfra on Intel Gaudi product
The following steps are optional. They're only required if you want to run the workloads on Intel Gaudi product.
-
Check the support matrix to make sure that environment meets the requirements.
-
Install and setup container runtime, based on the container runtime used by Kubernetes.
NOTE: Make sure you configure the appropriate container runtime based on the type of container runtime you installed during Kubernetes setup.
-
Install Intel Gaudi device plugin for Kubernetes.
Alternatively, Intel provides a base operator to manage the Gaudi software stack. Refer to this file for details.
Follow GMC README to install GMC into your kubernetes cluster. GenAIExamples contains several sample GenAI example use case pipelines such as ChatQnA, DocSum, etc. Once you have deployed GMC in your Kubernetes cluster, you can deploy any of the example pipelines by following its Readme file (e.g. Docsum).
To deploy GenAIExamples to Kubernetes using helm charts, you need Helm installed on your machine.
For a detailed version, see Deploy GenAIExample/GenAIComps using helm charts
You can use Terraform to create infrastructure to run OPEA applications on various cloud service provider (CSP) environments.