Skip to content

Latest commit

 

History

History
102 lines (75 loc) · 5.14 KB

deployment_guide.md

File metadata and controls

102 lines (75 loc) · 5.14 KB
Raw

Requirements

Before you deploy, you must have the following in place:

For prototyping, you need the following:

Step 1: Front-end deployment

  1. Clone and Fork this solution repository. If you haven't configured Amplify before, configure the Amplify CLI in your terminal as follows:
amplify configure
  1. In a terminal from the project root directory, enter the following command selecting the IAM user of the AWS Account you will deploy this application from. (accept all defaults):
amplify init
  1. Deploy the resourse to your AWS Account using the command:
amplify push
  1. After the Amplify deployment finishes, run the command bellow to obtain the Amazon S3 Bucket Amplify created. This information will be used later as a parameter in a clouformation
aws resourcegroupstaggingapi get-resources --tag-filters Key=user:Application,Values="COVID19L3NetAPP" Key=user:Stack,Values="dev" --resource-type-filters s3 --query 'ResourceTagMappingList[*].[ResourceARN]' --output text | grep -v deployment | awk -F':::' '{print $2}'
  1. Log into the AWS Management Console.
  2. Select AWS Amplify and select the COVID19L3NetApp
  3. At the Frontend environments tab connect to your github account poiting to the forked repo. More informatoin at https://docs.aws.amazon.com/amplify/latest/userguide/deploy-backend.html

Step 2: Back-end deployment

In this step we will execute three Cloudformation scripts:

  • cfn-vpc - This Cloudformation create the networking for the image creation EC2 instance, Lambda functions and EC2 instances that processes the model.
  • cfn-imageBuilder - It creates the EC2 Image Builder infrastructure that embeds the model into our custom AMI.
  • cfn-backend - Responsible for the creation of the underlying infrastrucutre of the solution. It includes the EC2 Auto Scaling configuration, SQS, VPC Endpoints, EFS and CloudFront

Step 2.1: VPC

  1. Log into the CloudFormation Management Console.
  2. Select Create stack with the With new resources option.
  3. Click Upload a template file, and then Choose file and select the cfn-vpc.yaml located at the /cfn directory of the repo
  4. Click Next.
  5. Give the Stack name a name (e.g. covid-19-app-vpc).

Step 2.2: EC2 Image Builder

  1. You also need the latest Deep Learning Amazon Machine Image (AMI) Id in the step. Please, run the command bellow to obtain it. Make sure run this command on the region you are executing the solution.
aws ec2 describe-images \
    --owners amazon \
    --filters 'Name=name,Values=Deep Learning Base AMI (Amazon Linux 2)*' 'Name=state,Values=available' \
    --query 'reverse(sort_by(Images, &CreationDate))[:1].ImageId' \
    --output text
  1. Log into the CloudFormation Management Console.
  2. Select Create stack with the With new resources option.
  3. Click Upload a template file, and then Choose file and select the cfn-ImageBuilder.yaml located at the /cfn directory of the repo
  4. Click Next.
  5. Give the Stack name a name (e.g. covid-19-app-ImageBuilder).
  6. Select a key-pair. If you don’t have any Amazon EC2 key-pair available create-your-key-pair, and repeat this step.
  7. On the AmazonLinuxAMI paste the AMI ID from the command listed at begining of thid step.

⚠️ Important Note: This step takes approximatelly 40min-60min as it spins up an instance and runs all the steps to create the AMI. Make sure it finishes succesfully to move to the next step

Step 2.3: Backend

  1. Log into the CloudFormation Management Console.
  2. Select Create stack with the With new resources option.
  3. Click Upload a template file, and then Choose file and select the cfn-backend.yaml located at the /cfn directory of the repo
  4. Click Next.
  5. Give the Stack name a name (e.g. covid-19-app-ImageBuilder).
  6. Select a key-pair that you have defined on Step 2.1 item 7.
  7. On the S3Bucket field past the bucket name obtained on the step 1.

Step 3: Lambda Function

3.1: Creating the Pydicom Layer

When a CT-Scan is submitted to be processed, a Lambda function is triggered to make sure that all files within the ZIP file are DICOM files. For this verification we leverage Pydicom. The first step to get this Lambda Function implemented is to create the Layer file.

  1. Go to the directory /backend/lambda and execute:
deploy.sh

This command launches a series of action that includes running a docker to retrieve Pydicom and create the layer file to be used on the lambda function.