How do I request a vCPU service quota increase for my EC2 On-Demand Instance?

Resolution

Check your usage and existing vCPU quotas

First, check the maximum number of vCPUs for each Amazon EC2 On-Demand Instance type.

Then, complete the following steps to check the number of vCPUs that each EC2 instance type currently uses:

1. Open the Amazon EC2 console.
2. In the navigation pane, choose Instance types.
3. Check the vCPUs column for your instance type.

Use these values to determine how many vCPUs to request based on need, potential traffic spikes, and failover requirements. It's a best practice to add the vCPUs that you use, plus the vCPUs that you plan to use, plus a buffer of 10-30%.

Calculate the number of vCPUs that you require

Calculate the quota required to launch new instances

To check whether you can launch an instance type that you aren't currently running, check the On-Demand Instance vCPU quota for that instance family.

For example, to launch 3 g4dn.xlarge instances, check the quota for G and VT instances. Each g4dn.xlarge instance uses 4 vCPUs for a total of 12 vCPUs. If your quota for G and VT instances is 8, then you can only run 2 g4dn.xlarge instances. To launch all 3 instances, submit a quota increase request to launch 12 vCPUs.

If you already have running instances in the G and VT instance families, then you must add their vCPUs to the total that you calculate. For example, you're running 3 g4dn.xlarge instances and want to also run 1 vt1.3xlarge instance. The 3 running instances use your entire quota of 12 vCPUs, and the vt1.3xlarge requires 12 vCPUs. To launch all the required instances, submit a quota increase request to run 24 vCPUs.

Calculate the quota for Standard instances

All On-Demand Standard instances for the A, C, D, H, I, M, R, T, and Z families have a shared quota. To calculate your vCPU requirements for these instance families, you must calculate the usage and needs across all Standard instances.

For example, you're running 64 vCPUs of Standard instances and you want to launch 2 t3.2xlarge and 2 m5.12xlarge instances. Your current vCPU quota to run On-Demand Standard instances is 64 vCPUs. Because both instance types that you want to add are Standard instances, you must combine their values to calculate the quota:

• t3.2xlarge instances use 8 vCPUs for each instance, so you need a total of 16 vCPUs for this instance type.
• m5.12xlarge instances use 48 vCPUs for each instance, so you need a total of 96 vCPUs for this instance type.

In total, you need an additional 112 vCPUs, plus the 64 vCPUs that you're already running. Submit a quota increase request of at least 178 vCPUs.

Calculate the quota to test across instance types

You can switch between instance types within a quota category without a new quota increase request as long as you stay below the applied quota. To evaluate different instance types within the same quota category, request vCPUs total to the largest test that you plan to run.

For example, you want to test P instances and your quota for On-Demand Instances in the P family is 64 vCPUs. You want to test a variety of instance sizes, such as the following example:

• 5 p5.4xlarge instances at 16 vCPUs for each instance, for a total of 80 vCPUs for this instance type.
• 2 p4dn.24xlarge at 96 vCPUs for the instance, for a total of 192 vCPUs for this instance type.

To keep your service quota increase request small and flexible, submit a vCPU request for a total 272 vCPUs. After you receive the quota increase, you can change the instance types across the P instance family. For example, replace the 2 p4dn.24xlarge instances with 1 p6-b200.48xlarge that also requires 192 vCPUs.

Note: In this scenario, specify the instance types in the Use case description to streamline the request process. Also, specify whether you plan to use the different instances in parallel or one at a time.

Submit a quota increase

Use the Service Quotas console to request an increase of the vCPU quota.

Important: Amazon EC2 service quotas apply to only one AWS Region. You must submit a service quota increase request for each Region that requires more resources.

Related information

Amazon EC2 service quotas

Requesting a quota increase

Amazon EC2 FAQs