Does AWS guarantee Fargate on-demand capacity availability?

Question

Hello,

I’m running an ECS service using the AWS Fargate (on-demand) launch type.

Recently, I encountered this event message when trying to launch a task: "was unable to place a task. Reason: Capacity is unavailable at this time. Please try again later or in a different availability zone."

Here are the details of my setup:
- I’m using Fargate on-demand, not Fargate Spot.
- The service is deployed in a single Availability Zone.
- At the moment of the error, I had no other Fargate tasks running in my account, so this was not related to vCPU or service quota limits.

This makes me wonder — Does AWS guarantee immediate Fargate on-demand capacity availability per Availability Zone?
If not, is there any official AWS documentation or SLA stating that Fargate on-demand capacity is best-effort and can be temporarily unavailable and what options exist to make Fargate capacity “always available”?

I’d like to understand the expected behavior and how AWS defines the reliability of Fargate capacity for on-demand workloads.

re:Post Agent · Answer

Based on the information available, AWS does not provide an explicit guarantee of immediate Fargate on-demand capacity availability in all circumstances.

While Fargate on-demand is designed to provide capacity when needed without requiring you to manage servers, there can be situations where capacity might be temporarily unavailable in a specific Availability Zone. This is different from Fargate Spot, which explicitly comes with the understanding that capacity might not be available at all times.

The error message you encountered suggests that even with Fargate on-demand, there can be temporary capacity constraints in specific Availability Zones. To improve reliability and availability of your workloads, consider these best practices:

1. Deploy your services across multiple Availability Zones to increase resilience against capacity constraints in any single zone.

2. For critical workloads that require high availability, consider using a combination of Fargate on-demand across multiple Availability Zones.

3. For less critical workloads, you might consider using a mix of Fargate on-demand and Fargate Spot to optimize costs while maintaining service availability.

Unlike Fargate Spot, which is explicitly subject to interruption with a two-minute warning during periods of high demand, Fargate on-demand is designed to be more reliable but does not come with an absolute guarantee of immediate capacity availability in every circumstance.

AWS Fargate is still a serverless compute engine that shifts infrastructure management to AWS, allowing you to focus on your applications rather than server management, but like any cloud service, it operates within the constraints of available physical resources in each region and zone.  
**Sources**  
[Amazon Amazon ECS launch types and capacity providers - Amazon Elastic Container Service](https://docs.aws.amazon.com/AmazonECS/latest/developerguide/capacity-launch-type-comparison.html)  
[Serverless Compute Engine – AWS Fargate Features – Amazon Web Services](https://aws.amazon.com/fargate/features/)  
[Using Amazon EC2 Spot and FARGATE_SPOT - Amazon Elastic Container Service](https://docs.aws.amazon.com/AmazonECS/latest/bestpracticesguide/ec2-and-fargate-spot.html)  
[Cluster capacity - Amazon Elastic Container Service](https://docs.aws.amazon.com/AmazonECS/latest/bestpracticesguide/capacity-cluster.html)

Does AWS guarantee Fargate on-demand capacity availability?

Add your answer

Relevant content