Skip to content

Latest commit

 

History

History
80 lines (51 loc) · 4.46 KB

README.md

File metadata and controls

80 lines (51 loc) · 4.46 KB
Raw

Use Terraform and Gitlab CI/CD to Build MLOps Pipeline on Amazon SageMaker

Introduction

Terraform is an “infrastructure as code” tool similar to AWS CloudFormation that allows you to create, update, and version your Amazon Web Services (AWS) infrastructure. Gitlab CI/CD is a tool for software development and enable you automatically build, test, deploy, and monitor your applications on AWS platform.

In this repository, Terraform and Gitlab CI/CD are used to deploy and orchestrate MLOps workflow on Amazon SageMaker platform.

Benefits

With this MLOps architecture, users can use AWS partners' tools to build an ML CI/CD pipeline on Amazon SageMaker. This pipeline enables customer rapid delivery and configurable workflow. The automated pipeline accelerates ML model production and can reduce delivery time from months to days.

Overview of Solution

The MLOps architecture is presented below.

The diagram shows Terraform provisions AWS sources that include lambda functions, and CloudWatch Events. A lambda function will create two projects and repositories in Gitlab. Each Gitlab project includes a CI/CD pipeline, which orchestrate ML workflow on Amazon SageMaker via AWS CLI.

Prerequisites

The following steps need to be complete before creating Gitlab projects.

  1. Create a token in Gitlab that will be used by the Lambda function defined in lambda_functions/lambda-seedcode-checkin-Gitlab to create 2 repositories in Gitlab populated with seed code for model building and model deployment.
  2. Create an IAM user for Gitlab. This user will be used by Gitlab Pipelines when interacting with SageMaker APIs. Take note of the AWS Access Key and Secret Key. This will be used in subsequent steps.
  3. Terraform CLI installed
  4. AWS CLI installed

Instruction

Step 1: Pass parameters

Go to Terraform files folder to input your own parameters (such as AWS account ID, region, Gitlab repo name, S3 bucket name. etc.) in:
variable.tf

Step 2: Create secret and load seed files to S3

Run ./init.sh (You may need to run chmod +x init.sh first to get execution permission)

It will create zip files for the seedcode and Lambda functions and upload them to S3. Pass command line arguments for the S3 bucket name, secret name, and secret token for Gitlab so that the Secrets Manager key can be created. The zip files created are:
1. zip_files/lambda-Gitlab-pipeline-trigger.zip
2. zip_files/lambda-seedcode-checkin-Gitlab.zip
3. zip_files/mlops-Gitlab-project-seedcode-model-build.zip
4. zip_files/mlops-Gitlab-project-seedcode-model-deploy.zip

Step 3: Run Terraform code

Go to Terraform files folder, run the following commands in Terminal:
terraform init
terraform apply

Terraform will create lambda functions, IAM service role, and CloudWatch Events.

lambda-seedcode-checkin-Gitlab will create two repositories in Gitlab.

Each repository will have a Gitlab CI Pipeline associated with it that will run as soon as the project is created. The first run of each pipeline will fail because Gitlab does not have the AWS credentials.

For each repository, navigate to Settings -> CI/CD -> Variables. Create 2 new variables - AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY with the associated information for your Gitlab role.

Trigger the pipeline in the model build repository to start a SageMaker Pipeline execution to train your model.

Once the SageMaker Pipeline to train the model completes, a model will be added to the SageMaker Model Registry. If that model is approved (in this repo, model is automatically approved), the Gitlab Pipeline in the model deploy repository will start and the model deployment process will begin. A SageMaker Endpoint will be created with the suffix -staging. A manual step in the Gitlab Pipeline is present to create an endpoint with the suffix -production.

Cleanup

  1. Log into your AWS Account, delete the endpoints, the secret, and zip files from the S3 bucket.
  2. Log into Gitlab account, delete the two projects created by the lambda function.
  3. Go to Terraform files folder, run the following commands in Terminal:
    terraform destroy

Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved. SPDX-License-Identifier: MIT-0