Amazon ECS clusters
An Amazon ECS cluster is a logical grouping of tasks or services that provides the infrastructure capacity for your containerized applications. When creating a cluster, you choose from the three primary infrastructure types, each optimized for different use cases and operational requirements.
Choosing the right cluster type
Amazon ECS offers three infrastructure types for your clusters. Choose the type that best matches your workload requirements, operational preferences, and cost optimization goals:
- Amazon ECS Managed Instances (Recommended)
-
Best for most workloads - AWS fully manages the underlying Amazon EC2 instances, including provisioning, patching, and scaling. This option provides the optimal balance of performance, cost-effectiveness, and operational simplicity.
Use when:
You want AWS to handle infrastructure management
You need cost-effective compute with automatic optimization
You want to focus on your applications rather than infrastructure
You need predictable performance with flexible scaling
- Fargate
-
Serverless compute - Pay only for the resources your tasks use without managing any infrastructure. Ideal for variable workloads and getting started quickly.
Use when:
You want completely serverless operations
You have unpredictable or variable workloads
You want to minimize operational overhead
You need rapid deployment and scaling
- Amazon EC2 instances
-
Full control - You manage the underlying Amazon EC2 instances directly, including instance selection, configuration, and maintenance.
Use when:
You need specific instance types or configurations
You have existing Amazon EC2 infrastructure to leverage
You require custom AMIs or specialized software
You need maximum control over the underlying infrastructure
Note
Amazon ECS Managed Instances is the recommended choice for most new workloads as it provides the best combination of performance, cost optimization, and operational simplicity while allowing AWS to handle infrastructure management tasks.
Cluster components
In addition to the infrastructure capacity, a cluster consists of the following resources:
-
The network (VPC and subnet) where your tasks and services run
When you use Amazon ECS Managed Instances or Amazon EC2 instances for the capacity, the subnet can be in Availability Zones, Local Zones, Wavelength Zones, or AWS Outposts.
-
An optional namespace
A namespace is used for service-to-service communication with Service Connect.
-
A monitoring option
CloudWatch Container Insights comes at an additional cost and is a fully managed service. It automatically collects, aggregates, and summarizes Amazon ECS metrics and logs.
Cluster concepts
The following are general concepts about Amazon ECS clusters.
-
You create clusters to separate your resources.
-
Clusters are AWS Region specific.
-
Clusters can be in any of the following states.
- ACTIVE
-
The cluster is ready to accept tasks and, if applicable, you can register container instances with the cluster.
- PROVISIONING
-
The cluster has capacity providers associated with it and the resources needed for the capacity provider are being created.
- DEPROVISIONING
-
The cluster has capacity providers associated with it and the resources needed for the capacity provider are being deleted.
- FAILED
-
The cluster has capacity providers associated with it and the resources needed for the capacity provider have failed to create.
- INACTIVE
-
The cluster has been deleted. Clusters with an
INACTIVEstatus may remain discoverable in your account for a period of time. This behavior is subject to change in the future, so make sure you do not rely onINACTIVEclusters persisting.
-
A cluster can contain a mix of tasks that are hosted on Amazon ECS Managed Instances, AWS Fargate, Amazon EC2 instances, or external instances. Tasks can run on Amazon ECS Managed Instances, Fargate or EC2 infrastructure as a launch type or a capacity provider strategy. If you use EC2 capacity providers, Amazon ECS doesn't track and scale the capacity of Amazon EC2 Auto Scaling groups.
-
A cluster can contain a mix of Amazon ECS Managed Instances capacity providers, Auto Scaling group capacity providers, and Fargate capacity providers. A capacity provider strategy can only contain Amazon ECS Managed Instances capacity providers, Auto Scaling group capacity providers, or Fargate capacity providers.
-
You can use different instance types for the Amazon ECS Managed Instances and EC2, or Auto Scaling group capacity providers. An instance can only be registered to one cluster at a time.
-
You can restrict access to clusters by creating custom IAM policies. For information, see Amazon ECS cluster examples section in Identity-based policy examples for Amazon Elastic Container Service.
-
You can use Service Auto Scaling to scale Fargate tasks. For more information, see Automatically scale your Amazon ECS service.
-
You can configure a default Service Connect namespace for a cluster. After you set a default Service Connect namespace, any new services created in the cluster can be added as client services in the namespace by turning on Service Connect. No additional configuration is required. For more information, see Use Service Connect to connect Amazon ECS services with short names.
Capacity providers
Amazon ECS capacity providers manage the scaling of infrastructure for tasks in your clusters. Each cluster can have one or more capacity providers and an optional capacity provider strategy. You can assign a default capacity provider strategy to the cluster. The capacity provider strategy determines how the tasks are spread across the cluster's capacity providers. When you run a standalone task or create a service, you either use the cluster's default capacity provider strategy or a capacity provider strategy that overrides the default one. The cluster's default capacity provider strategy only applies when you don't specify a launch type, or capacity provider strategy for your task or service. If you provide either of these parameters, the default strategy isn't used.
Amazon ECS offers three types of capacity providers for your clusters:
- Amazon ECS Managed Instances capacity providers
-
AWS fully manages the underlying Amazon EC2 instances, including provisioning, patching, scaling, and lifecycle management. This provides the optimal balance of performance, cost-effectiveness, and operational simplicity. Amazon ECS Managed Instances capacity providers automatically optimize instance selection and scaling based on your workload requirements.
With Amazon ECS Managed Instances, you benefit from:
Automatic instance provisioning and scaling
Managed patching and security updates
Cost optimization through intelligent instance selection
Reduced operational overhead
- Fargate capacity providers
-
Serverless compute where you pay only for the resources your tasks use without managing any infrastructure. You just need to associate the pre-defined capacity providers (Fargate and Fargate Spot) with the cluster.
- Auto Scaling group capacity providers
-
When you use Amazon EC2 instances for your capacity, you use Auto Scaling group to manage the Amazon EC2 instances. Auto Scaling helps ensure that you have the correct number of Amazon EC2 instances available to handle the application load. You have full control over the underlying infrastructure.
A cluster can contain a mix of tasks that are hosted on Amazon ECS Managed Instances, AWS Fargate, Amazon EC2 instances, or external instances. Tasks can run on Amazon ECS Managed Instances, Fargate or EC2 infrastructure as a launch type or a capacity provider strategy. If you use EC2 as a launch type, Amazon ECS doesn't track and scale the capacity of Amazon EC2 Auto Scaling groups.
A cluster can contain a mix of Amazon ECS Managed Instances capacity providers, Auto Scaling group capacity providers, and Fargate capacity providers. A capacity provider strategy can only contain Amazon ECS Managed Instances capacity providers, Auto Scaling group capacity providers, or Fargate capacity providers.
