CSUN-RCV Terraform Setup

This repository contains Terraform code to set up AWS resources supporting the California State University Northridge Real-time Computer Vision system.

see also:

• csun-rcv
• csun-rcv-ui-demo

Prerequisites

Terraform v1.6.

Container Images

The following container images must be uploaded to Amazon ECR. Build scripts are available in the linked repositories:

• ocv-dnn-detector-lambda
• csun-rcv

Commands

1. Run terraform plan -out tfplan to generate an execution plan.
2. Run terraform apply tfplan to apply the generated plan and create the resources.

These commands should be executed from the project's root directory.

Modules

The Terraform setup includes the following modules, each corresponding to a specific service:

• dynamoDB: DynamoDB table setup.
• integration: Fargate task for CSUN-RCV image. Launches the task when the Kinesis Video Stream becomes active and automatically shuts down when the stream ends.
• kvs: Kinesis Video Stream setup.
• lambda: Object detector Lambda setup.
• network: Network configuration.
• s3: S3 bucket setup.
• ssm: Systems Manager setup.

API Endpoint

The API URL for the CSUN-RCV UI demo can be found through the AWS Management Console in the API Gateway section.

Outputs

After Terraform applies the configuration, you can find the following outputs:

• lambda_function_name: Name of the AWS Lambda function for object detection.
• dynamodb_table: Name of the DynamoDB table.
• s3_bucket: Name of the S3 bucket for storing processed frames.
• kvs_stream: Name of the Kinesis Video Stream.
• ssm_parameter: Path to the AWS Systems Manager parameter containing system metadata.

Tear Down

To tear down all resources created by Terraform, follow these steps:

1. Run terraform destroy from the project's root directory.
2. Confirm by typing yes when prompted.

Cost Considerations

Lambda (Object Detector)

• 1024 MB memory configuration.
• Lambda Pricing

Fargate (Integration Module)

• CPU: 1024, Memory: 2048 MB configuration.
• Fargate Pricing

Kinesis Video Stream

• Incur charges based on usage.
• KVS pricing

S3

• Fragments saved to S3 are discarded after 24 hours.
• May incur charges based on storage and request usage.
• S3 pricing

Cost Structure

• The system incurs no charges if not running.
• When running, Lambda first falls within the free tier before incurring charges.
• Kinesis Video Stream and S3 may incur charges based on usage.
• ECS Fargate incurs charges only when the task is running.

Monitor usage and costs regularly to optimize spending and ensure cost efficiency.

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.