Deployment

Before using Kernel Memory in Azure, it is important to understand its architecture. You can find the architecture guide here.

For the opinionated deployment of Kernel Memory, follow the instructions in the infra README.

To deploy, use the “Deploy to Azure” button or run the following command in the CLI:

az bicep build -f main.bicep

It’s important to understand that Azure resource usage is billed based on the resources you use. Refer to the cost section to understand the costs associated with Kernel Memory.

Kernel Memory takes about 5 minutes to deploy and start using. To reduce costs while experimenting with Kernel Memory, delete resources after use. Note that all uploaded data will be lost upon deletion.

Deployment Customization

Infrastructure as Code (IaC) is a key part of Kernel Memory deployment. You can customize the deployment by modifying the .bicep files in the Kernel Memory GitHub repository. If Bicep or IaC is new to you, learn more about it in the Fundamentals of Bicep.

Considerations </br>

Do Not Over-Provisioning Resources: Deploying larger or more resources than necessary, leads to increased costs without corresponding benefits. Start with smaller instances and scale up based on actual usage.

Set-up Budgets and Alerts Failing to set up budgets and alerts to monitor and control spending. Solution: Use Azure Cost Management tools to track expenses and set alerts for unusual spending patterns.

Trouble shooting Deployment fails are often a result of missing permissions. Ensure that the Managed Identity has the necessary permissions to access Azure resources.