Demystifying Google Committed Use Discounts

Demystifying Google Committed Use Discounts (CUDs) takes a little time, but it can be advantageous to better understand Google’s discount system, what it applies to, and when best to use it. Is the effort worthwhile when compared to Amazon Web Services (AWS) or Microsoft Azure’s discounting programs? For many businesses, it could be. 

Google has offered CUDs for several years— a GCP discount system designed to rival AWS Reserved Instances, but with greater flexibility and no advanced payments required to take advantage of the maximum available discount. 

In certain scenarios, Google’s discount system results in there being little difference between the prices of AWS Reserved Instances and Google Cloud instances for businesses prepared to commit to a lengthy contract period. However, there are other pros and cons of Google CUDs. 

So, what makes Google Committed Use Discounts a better system for some businesses and not for others? Or are the businesses not yet taking advantage of Google´s discount system failing to understand the benefits? Let’s take a deeper look at CUDs to find out more, including how they differ for Google Compute Engine and other GCP services, as well as how to avoid confusion from Google’s offering of Sustained Use Discounts. 

How do Google Committed Use Discounts work?

If your business has stable and predictable workloads that are likely to remain the same for more than a year, you can take advantage of Google CUDs by simply committing to a specific number of resources or money spent for a certain amount of time. These discounts have some flexibility and can cover a variety of resources offered by Google.  

Once you have committed to a contract, you are billed each month for the duration of your preferred term, and you pay for the services whether you use them or not. Any services you use above the commitment level arebilled at the on-demand rate but may qualify for Sustained Use Discounts, automatic discounts on incremental usage triggered by running specific resources for a significant portion of a billing month. The commitment required by a buyer can be based on a specific number of resources or amount of spend. 

To commit to a level of service with Google, you must have sufficient space in your quota, which restricts the amount of a particular shared resource that you can use. Rate Quotas limit the number of requests you can make to an API or service, while Allocation Quotas restrict the use of resources that don’t have a rate of usage. Quotas are enforced by Google to prevent unforeseen spikes in usage and overloaded services, and to help with your own resource management. Quota increases can be requested, at which point they are evaluated by Google. 

Resource-based commitments  

Resource-based commitments offer a discount for your commitment to using a minimum-level of resources in a region. Resource-based CUDs are only available for Google Compute Engine. 

Google Compute Engine CUDs  

• Up to a 57% discount for machine types or GPUs 
• Up to a 70% discount for memory-optimized machine types 
• Apply to vCPUs, memory, GPUs, and local SSDs within a region, which can be shared across projects on the same Cloud Billing account 
• Can be purchased for a wide variety of resource types including E2, N2, N2D, T2D, N1, and A2 
• Cannot be stacked for burst scenarios 

The ability to share Committed Use Discounts across projects on the same Cloud Billing account can be easily tracked using Attribution in Google’s cost management interfaces. To do this, Discount Sharing must first be enabled. Once enabled, Google offers two different attribution methods to track which projects are responsible for your usage and benefiting from CUDs. Without Discount Sharing, CUDs will live in projects and only apply to usage in the project the CUD was purchased in. 

1. Proportional attribution (default): Applies discounts directly in proportion to the amount of total eligible usage consumed by each project.  
2. Prioritized attribution: Applies discounts based on a distribution that you specify for specific projects.  

If desired, you can mix these two methods by prioritizing certain projects and then specifying discounts proportionally to any remaining projects.  

The restrictions on GCE Committed Use Discounts make it more difficult for some businesses to take advantage of the system. In these cases, they may turn to AWS Reserved Instances and Azure Reserved VM Instances for their needs, which both offer more flexibility with selling back (AWS) or receiving prorated refunds (Azure) on reservations.  

Comprehensive information on GCE CUDs can be found here from Google. 

Spend-based commitments 

For spend-based commitments, you will receive a discount for committing to a minimum amount of spend for a product or service offered by Google. Spend-based Committed Use Discounts are available for Cloud SQL, Google Cloud VMware Engine, Cloud Run, and Google Kubernetes Engine.  

Cloud SQL CUDs  

• One-year commitment: 25% discount 
• Three-year commitment: 52% discount 
• Apply to aggregate CPU and memory usage in a region, except for shared CPU machine types 
• Do not apply to storage, backups, IP addresses, network egress, or licensing 

Google Cloud VMWare Engine CUDs 

• One-year commitment: 24.4% discount if invoiced monthly, 30% if paid in full at start of contract 
• Three-year commitment: 42.5% discount if invoiced monthly, 50% if paid in full at start of contract 
• Apply to aggregate VMware Engine nodes in a region 
• Do not apply to storage, backups, IP addresses, network egress, or licensing 

Cloud Run CUDs 

• One-year commitment: 17% discount 
• Apply to all aggregated Cloud Run CPU, memory, and request usage in a region 
• Do not apply to networking charges  
• Cloud Run CUDs are measured in dollars per hourly on-demand commitment

Google Kubernetes Engine (Autopilot Mode) CUDs 

• One-year commitment: 20% discount 
• Three-year commitment: 45% discount 
• Apply to aggregate GKE Autopilot Pod workload CPU, memory, and ephemeral storage in a region 
• Do not apply to the cluster management fee, Spot Pods, or to GKE Standard mode compute nodes 
• GKE (Autopilot Mode) CUDs are measured in dollars per hourly on-demand commitment 

The advantages of Google Committed Use Discounts

The most significant advantage is that no prepayment is required. With both AWS and Azure Reserved Instances, you must pay upfront to benefit from the maximum discounts. If you only partially pay upfront or on a monthly cycle, you won’t be able to take full advantage of the discounts offered by AWS or Azure. You can learn more about comparing the services of the big three cloud providers (AWS, Azure, GCP) in our eBook here. 

The second advantage is the fact you are committed to a specific number of vCPUs and memoryindependently, rather than machine types. Combined with custom instances, this gives you maximum flexibility to adjust the instance CPU and RAM independently, a feature unique to Google. 

There are many advantages on Google Cloud related to the network, disk, GPU and CPU characteristics that its two major competitors cannot match. Don´t forget that Google Cloud also offers the option of creating custom machines, facilitating still more flexibility in your deployments. 

Avoiding Confusion with Sustained Use Discounts 

Committed Use Discounts can often be confused with Sustained Use Discounts, which are an entirely different discount category offered by Google. Sustained Use Discounts are automatic discounts you receive on incremental usage after running Google Compute Engine resources for a significant portion of a billing month. The longer you continuously run your resources in a given billing month, the greater your potential discounts on incremental usage.  

Google will automatically calculate and apply Sustained Use Discounts based on your usage and will include these discounts automatically inline on your bill and at the end of the billing cycle as a separate line item.Remember that these discounts will only apply to incremental usage after you have already achieved a required level of sustained use in a billing month.