Atolio Deployment

This document describes the steps required to deploy Atolio into your cloud environment.

Cloud Deployment

For high-level architecture (including cloud platform services that are used), view the following diagram:

Deployment Process

The following steps outline the process for deploying Atolio to your cloud environment:

  1. Complete the necessary deployment prerequisites for your desired cloud deployment:
  1. Setup authentication via your SSO provider - see Configuring Authentication.
  2. Deploy the Atolio cloud infrastructure, either self-service or guided by your Atolio support team, using the scripts provided in the Atolio GitHub repository.
  3. Configure your connectors - see Configuring Sources.

Note: Depending on your scale requirements, Atolio may recommend an additional pre-processing step to scan and inventory the content of an expected connector. This may result in configuration changes (related to Vespa node/cluster size) in step 3.

Additional Resources

Operations & Maintenance

The Atolio GitHub repository contains:

  • Deployment documentation
  • Terraform for the Atolio stack infrastructure
  • Configuration files for Atolio services
  • Maintenance scripts

After deployment prerequisites, the initial infrastructure deployment will typically take up to 90min with your Atolio support team. Additional time can be scheduled for help with connector configuration, upgrades, or other operations.

Connector Extensibility

Atolio provides the ability to author custom connectors based on a separate Go Connector SDK. Reach out to your Atolio support team for access to the SDK repository and Best Practices Guide. We will add additional SDKs for supporting custom connectors in the future.

Model Extensibility

Atolio supports a variety of options for connecting large language models (LLMs) from services such as OpenAI, Azure OpenAI, and Amazon Bedrock. Additional configurations will be supported in the future.