How to Get Started with AWS Fargate: A Complete Guide Tutorial and Next Steps

Setting Up AWS Fargate

AWS Fargate has a bit of a learning curve, but there are a number of AWS Fargate tutorials (e.g. “Monolith to Microservices with Docker and AWS Fargate”) to help you get started. If you prefer to leap right in, then follow the steps in this section.

To start using AWS Fargate, you should be familiar with terminology such as “containers” and “tasks”:

• Containers are packages of source code together with the required libraries, frameworks, dependencies, and configurations. This helps applications run predictably and reliably, no matter the environment.
• Tasks are “blueprints” that define which containers should run together in order to form a software application.
• Clusters are logical groupings of one or more tasks.

Read through the first-run experience for AWS Fargate, which walks you through the process of creating Fargate containers, tasks, and clusters. Under the heading “Container definition,” select one of the default images (e.g. “sample-app”) and review the task definition. Click “Next” at the bottom of the page.

Now that you’ve determined the container and task, you need to define your service, which sets the number of simultaneous instances of a task definition within a single cluster. If desired, you can select the “Application Load Balancer” option, which helps redirect traffic during times of heavy demand (although it’s not necessary for this AWS Fargate tutorial).

Click “Next” to start configuring your cluster. In this Fargate tutorial, the only option you can change is the name of the cluster (which is already set to “default”). Finally, click “Next” to review your settings, and click “Create” to launch the service.

Once it launches, click on “View service” to view the cluster in your Amazon ECS dashboard. From here, you can click on “Update” to modify the cluster settings, or “Delete” to finish the AWS Fargate tutorial and start using the service for real serverless workloads.

Migrating to AWS Fargate

You may already have been using containers elsewhere on AWS (e.g. Amazon ECS or EKS)—but now you want to get started enjoying AWS Fargate’s serverless capabilities. The good news is that migrating to AWS Fargate isn’t too difficult.

Fargate is intended to work with a container orchestration service such as Amazon ECS (Elastic Container Service) or Amazon EKS (Elastic Kubernetes Service). As a nice result of this compatibility, you can use Fargate with the ECS console, since both Fargate and ECS use the same API actions.

Amazon provides a tutorial for migrating your Amazon ECS containers to AWS Fargate. The basic steps of this migration are as follows:

• In your task definition parameters, the setting “network mode” needs to be changed from bridge, which is the default for ECS, to awsvpc. This mode gives each of your Fargate tasks individual elastic network interfaces and IP addresses so that multiple applications can use the same port number.
• As a consequence of using awsvpc, AWS Fargate users need an Identity and Access Management (IAM) role called AWSServiceRoleForECS.
• Task-level resources need a defined amount of CPU (specified either in number or in vCPUs) and memory (specified in gigabytes or megabytes).

These are just some of the most important steps to perform an AWS Fargate migration; for the full list of possible changes, check out Amazon’s guide “Migrating Your Amazon ECS Containers to AWS Fargate.” You can also look at the Task Definitions for Amazon ECS GitHub repository, which contains multiple example task definitions for AWS Fargate.