How do I request an EC2 vCPU limit increase for my On-Demand Instance?

Short description

EC2 vCPU limit increases are submitted as a vCPU value. To request an increase, first determine how many vCPUs that your On-Demand Instances are using. You can use the vCPU limits calculator to measure the number of vCPUs that you are currently using against vCPU-based limits to determine the appropriate service limit increase to request. You can create a service limit request directly from the vCPU limits calculator.

Resolution

Request an On-Demand Instance limit increase

1. Sign in to the Amazon EC2 console.
2. In the Region selector, choose the AWS Region where you want to request a vCPU increase.
3. (Optional) To view the instances that are already running, from the EC2 Dashboard, under Resources, choose Instances (running).
4. In the navigation pane, Choose Limits.
5. Choose Calculate vCPU limit to access the vCPU limits calculator.
6. In the Limits calculator page, choose Add instance type.
7. For Instance type, select an available instance type from the dropdown list.
8. For Instance count, enter the number of new instances (of the type that you chose in step 6) that you want.
9. Continue adding instance types as needed.
10. Review the Limits calculation section. If the New limit exceeds the Current limit, choose Request on-demand limit increase to submit a service limit increase request. The value in the New limit field is the number of vCPUs that you must request.

Example calculations

Note: You can use the vCPU limits calculator for these calculations.

Scenario 1: You're not currently running any G family instances. You want to launch three g4dn.xlarge instances.

Your current vCPU limit is Running On-Demand All G instances = 8 vCPU.

A g4dn.xlarge instance has a footprint of 4 vCPUs. Your account can already run two g4dn.xlarge instances, based on your G instances limit of 8.

In this scenario, submit a vCPU service limit increase request for 12 vCPUs (3*4).

Format your request as follows:

Limit type: EC2 Instances
Region: US West
Primary Instance Type: Running On-Demand All G instances
New limit value: 12

Note: Replace US West with the Region where you're requesting a limit increase.

Scenario 2: You're currently running 64 vCPUs worth of instances. You want to launch two additional m5.2xlarge and two additional m4.10xlarge instance types.

Your current vCPU limit is Running On-Demand All Standard (A, C, D, H, I, M, R, T, Z) instances = 64 vCPU.

Both instance types that you want to launch are included in the "All Standard (A, C, D, H, I, M, R, T, Z) instances" limit. Usage for these instance types is added together and governed by a single limit. The instance-to-vCPU conversion rates for these two types of instances are as follows:

• *.2xlarge instances = 8 vCPU per instance
• *.10xlarge instances = 40 vCPU per instance.

Therefore, the additional number of vCPUs needed is calculated as follows:

[2 * 8] + [2 * 40] = 96.

You need 96 additional vCPUs if your account is already running 64 vCPUs worth of instances.

In this scenario, submit a vCPU service limit increase request for 160 vCPUs (64 + 96) or higher.

Format your request as follows:

Limit type: EC2 Instances
Region: US West
Primary Instance Type: Running On-Demand All Standard (A, C, D, H, I, M, R, T, Z) instances
New limit value: 160

Note: Replace US West with the Region you're requesting a limit increase in.

Scenario 3: You want to test more powerful instance types, but aren't sure which instances you need. You want to play around before deciding. Therefore, your service limit increase request needs to provide you flexibility.

Your current vCPU limit is Running On-Demand All P instances = 64 vCPU.

The vCPUs of an instance within a family can range from 1 to 128 vCPUs, depending on the size. vCPU limits give you the versatility to start with one instance type, and then switch to another instance type without submitting a new limit increase, as long as you stay within your vCPU limit.

For example, you might request a combination of P instance types:

• Five p3.2xlarge instances, 8 vCPU per instance = 40 vCPU
• One of the largest P-Family instance type, the p3dn.24xlarge, 96 vCPU per instance = 96 vCPU

In this scenario, to keep your limit request small while retaining flexibility, submit a request for 136 vCPU (40 + 96).

Format your request as follows:

Limit type: EC2 Instances
Region: US West
Primary Instance Type: Running On-Demand All P instances
New limit value: 136

Note: Replace US West with the Region that you're requesting a limit increase in.

To streamline the processing of your limit increase request, in the Use case description section of the request form, include specific instance types (for example, p3.8xlarge or p3dn.24xlarge). Also, note whether you're expecting to use the different instances in parallel or one at a time.

Related information

On-Demand Instances

Troubleshooting instance launch issues