Alain Airom

Posted on Mar 17

Deploying rapidly an enterprise grade Generative AI platform

#watsonx #genai #governance #ibmcloud

Using deployable architectures on IBM Cloud to provision and use generative AI platforms.

Introduction

1- What are deployable architectures?

Creating secure, compliant, and scalable application infrastructure can be difficult to set up and costly to maintain. Instead of figuring out how to assemble a compliant infrastructure architecture on your own, you can take advantage of modules and deployable architectures. Modules and deployable architectures can help you to create a framework around how resources are deployed in your organization’s accounts. By working with these reusable configurations, you can define the standard for deployment once and ensure that it is easily repeatable for each member of your organization.

For example, think about an architect who is building an apartment complex. These designs are typically executed in a modular way. There are patterns for standard one-bedroom, two, or three-bedroom apartments. The builder can combine the standard apartments, each functional in their own way, into a larger, more complex, but functional living arrangement. IBM applied this same analogy to deploying solutions on the cloud. Rather than your organization spending months figuring out how to get services and software to work together, you can use IBM Cloud’s well-architected patterns. Each pattern is packaged as composable, automated building blocks known as modules and deployable architectures.

2- What is a module?

A module is a stand-alone unit of automation code that can be reused by developers and shared as part of a larger system. Similar to Node.js or Python packages, modules are a convenience to developers who are managing related resources. While it is possible to use modules alone, they’re more powerful when you combine them to build a deployable architecture. Modules that are created by IBM Cloud are made available in the IBM Terraform modules public GitHub org. For example, the Red Hat® OpenShift® VPC cluster on IBM Cloud module installs and configures a Red Hat OpenShift cluster on IBM Cloud.

TLDR..

What is watsonx.governance

The watsonx.governance™ toolkit seamlessly integrates with your existing systems to automate and accelerate responsible AI workflows to help save time, reduce costs and comply with regulations. You also benefit from the ability to deploy AI wherever it makes sense across your hybrid cloud.

What is watsonx.ai

IBM watsonx.ai™ is an enterprise-grade studio for developing AI services and deploying them into your applications of choice with a collection of the APIs, tools, models and runtimes you need to turn your ideas and requirements into reality.

Use-case: watsonx.ai SaaS with Assistant and Governance

The deployable architecture is accessible directly here: https://cloud.ibm.com/catalog/architecture/deploy-arch-ibm-watsonx-ai-saas-e8ad6597-8c1a-466a-8bb7-243a109daaa8-global?catalog_query=aHR0cHM6Ly9jbG91ZC5pYm0uY29tL2NhdGFsb2cjZGVwbG95YWJsZV9hcmNoaXRlY3R1cmVfdGFi

Following the introduction and knowing what a deployable architecture is, let us dig into the watsonx.ai (Assistant and Governance) and learn how easy the solution could be deployed to provide AI across a business, managing all data sources, and accelerating responsible AI workflows on one platform.

The watsonx.ai SaaS with Assistant and Governance deployable architecture automates the deployment and setup of the watsonx platform in an IBM Cloud account. The IBM watsonx platform is made of several services working together to offer AI capabilities to end users, who can explore them using IBM watsonx projects. The automation also configures a starterwatsonx project for an existing IBM Cloud user in the target IBM Cloud account.

The deployable architecture can help set up an IBM watsonx platform that’s ready to use in one or more Enterprise accounts. It grants administrator access to an AI Researcher or any other professional who needs access to advanced AI technologies for their work.

The deployable architecture can also be used as part of a larger solution, where it is included in a stack of other components. To facilitate those business challenges, the watsonx.ai SaaS with Assistant and Governance deployable architecture provides output parameters that can be used programmatically for wiring the deployable architecture to the other components of the stack, and it implements the flexibility to install additional Watson services.

The watsonx.ai SaaS with Assistant and Governance deployable architecture performs the following steps to create a ready to use IBM watsonx platform in a target IBM Cloud account, resource group, and IBM Cloud location:

Creates or uses an existing resource group in the target IBM Cloud account
Creates the following services in the target resource group and location:
Cloud Object Storage
Watson Studio
Watson Machine Learning
Optionally, it can create one or more of the following services in the target resource group and location:
watsonx.data
watsonx.governance
watsonx Assistant
Watson Discovery
watsonx Orchestrate
It creates the IBM watsonx user profile for an existing user in the target IBM Cloud account. This user is also referred as IBM watsonx admin.
If you provided the CRN of a IBM Key Protect instance in the same target account and location of the watsonx services, then it enables storage delegation for the Cloud Object Storage instance.
It creates a starter IBM watsonx project.

And all this in a few minutes… ✌️

Architecture

The high-level architecture diagram of what is deployed is presented below, and helps visually understand what components are going to be deployed.

The second diagram gives a clear understanding of what comes out of the box from the deployment, and also visually clarifies what additional functionalities might be necessary to add.

Regarding the provisioning by itself, it relies on Terraform scripts which are available and accessible on public GitHub repository.

Another feature which is extremely important for a company is how much would this architecture cost! Well for that all is clear using the cost estimator which is provided.

Conclusion

Provisioning an enterprise grade generative AI platform could be complicated and in terms of determining which services to provision and what to provision. IBM provides a standardized packaged tool which implements the components and services in minutes, distilling the burden of complicated deployments.

Amplify your impact where it matters most — building exceptional apps.

Leave the infrastructure headaches to us, while you focus on pushing boundaries, realizing your vision, and making a lasting impression on your users.

Get Started

DEV Community