AWS ECS with Terraform
While we previously recommended AWS ECS Fargate as our primary hosting option for Sombra, we now recommend Kubernetes/Helm-based deployments for their cloud-agnostic support and enhanced capabilities. AWS Fargate does not support GPU nodes, making it unsuitable for deployments requiring the LLM Classifier.
That said, the ECS deployment option remains supported for environments where Kubernetes isn't available or practical. Be aware that our future infrastructure-as-code updates will prioritize the Helm chart implementation.
Prerequisites
Follow our Getting Started guide to get access to our container registry, and get all the environment variables you'll need when configuring Sombra.
Use the example ECS Terraform configuration in the module's example folder. Note that you may already have an existing VPC where you want to deploy Sombra, in which case you can omit the VPC module.
For a complete list of customizable fields, refer to the module's variables file.
Go to the Sombra Gateways page and click the "Test" button under "Test Gateway Connection" to make a GET request to the /test endpoint of the Sombra Gateway.
You should see a successful message if the connection is working. Congratulations on successfully deploying Sombra!
When using ECS for Sombra deployments, be aware of these limitations:
- No support for GPU-based workloads, such as the LLM Classifier
- Less flexibility for scaling and resource management
- AWS-specific deployment, making migration to other cloud providers more difficult