Cloud Comparison Financial Management

Comparing AWS Reserved Instances Vs. Google Committed Use Discounts

A price comparison matching AWS Reserved Instances vs. Google Committed Use Discounts is not a sound basis for choosing one Cloud Service provider above another. We explain why and provide examples of what you should look for when evaluating long-term discounts for cloud services.

AWS’ Reserved Instances and Google’s Committed Use Discounts both provide an opportunity to reduce cloud costs by committing to a level of service for a predetermined period. The two cloud service providers respectively advertise discounts of up to 75 percent and 57 percent on pay-as-you-go prices depending on the length of the commitment and—in the case of AWS—how much you pay upfront.

However, comparing AWS Reserved Instances vs. Google Committed Use Discounts on the percentage of the discounts does not give a clear picture of which option may be best suited to your needs—especially when the two company’s discount programs are frequently changing (for example, with the introduction of Convertible Reserved Instances, the extension of the Committed Use Discount Program, and the launch of AWS Savings Plans).

The difficulty in compiling an AWS RI vs Google CUD price comparison

If you’ve ever been tempted to compile a price comparison matching AWS Reserved Instances vs. Google Committed Use Discounts, you’ll likely appreciate the difficulty involved due to the number of variables. Although both discount programs are similar inasmuch as they offer discounts in return for one-year and three-year commitments, there are many ways that they differ. For example:

  • AWS offers the options to pay all, partially, or nothing upfront in return for a larger discount.
  • Google Cloud Platform offers the opportunity to apply discounts to customized instances.
  • AWS offers additional volume discounts to businesses that spend more than $500,000 a year.
  • Google’s discounts are restricted to instances with between 0.9GB and 6.5GB memory per CPU.
  • AWS discounts can be applied to Standard or Convertible Reserved Instances, or Savings Plans.
  • Google Committed Use Discounts also apply to GPUs, Cloud TPU Pods, and local SSDs.
  • The two cloud service providers charge differently for dedicated “sole-tenancy” instances.
  • AWS limits businesses to purchase 80 Reserved Instance purchases per month (20 per Availability Zone (AZ) plus 20 per Region).
  • There’s no maximum number of Google instances that qualify for a Committed Use Discount.
  • AWS offers Reserved Instances with guaranteed capacity when an AZ is specified.

What’s also worth pointing out is that once you have fully loaded an AWS Reserved Instance reservation, any remaining instances you deploy on AWS are charged at the full On-Demand rate. With Google Cloud Platform, any instances you run above your Committed Use commitment qualify for Sustained Use Discounts. Obviously, if you had a lot of On-Demand AWS instances running long-term, you’d purchase more Reserved Instances, but in the short-term, Google Cloud Platform may be the better option.

You can learn more about comparing the services of the big three cloud providers (AWS, Azure, GCP) in our eBook here.

AWS Reserved Instances vs. Google Committed Use Discounts comparison

In order to conduct a like-for-like AWS Reserved Instances vs. Google Committed Use Discounts comparison, we’ve selected regional, standard AWS instances and compared their Reserved Instance (RI) prices per month (without any upfront payments being applied) against similar Google instances after a Committed Use Discount (CUD) has been applied. Prices are expressed in U.S. dollars per month.

AWS Reserved Instances vs. Google Committed Use Discounts Comparison


AWS 1 Year RI 3 Years RI Google 1 Year CUD 3 Years CUD
m5.xlarge $88.33 $60.59 n1-standard-4 $87.38 $62.42
m5.2xlarge $176.66 $121.18 n1-standard-8 $174.75 $124.84
r5.xlarge $116.07 $79.57 n1-highmem-4 $108.81 $77.73
r5.24xlarge $2,781.30 $1,907.49 n1-highmem-96 $2,611.43 $1,865.56

To achieve the maximum cost savings, you need to have total visibility of your assets, an insight into how your instances are being used, and the right mechanism in place to manage instances across multiple clouds. Once you have total visibility, you first need to determine what the maximum capacity requirements are for each instance and provision them accordingly. For example:

If a four-core instance only requires 12GB of memory, it would be wasteful to deploy an m5.xlarge instance on AWS (monthly cost for a three-year Reserved Instance $60.59) or an n1-standard-4 instance on Google Cloud Platform (monthly cost after a three-year Committee Use Discount is applied $62.42), when the opportunity exists to deploy a customized instance on Google Cloud Platform for the monthly cost of $58.24.

Naturally, there are additional factors to take into account when mixing and matching AWS Reserved Instances and Google Committed Use Discounts—such as comparative storage costs and the fact you cannot modify, exchange, or sell Committed Use Discounts—but the saving of ~$2 per instance per month will likely be worth the effort of provisioning instances to match their capacity requirements.

Gaining insight into and managing Instances on different discount programs

Both AWS and Google Cloud Platform offer tools for gaining insight into and managing instances deployed on their own platforms, but not across multiple platforms simultaneously. In order to do this, you need a solution such as CloudHealth´s cloud management platform that enables you to analyze the performance and utilization of all your instances regardless of where they’re deployed.

CloudHealth will make recommendations about the most appropriate configuration of your instances so you can take advantage of discount programs effectively. The platform will continue to monitor utilization throughout the life of the Reserved Instance and make further recommendations based on historical usage, reservation types, and expiring commitments, so you can be sure you are always achieving the maximum possible cost reduction without sacrificing performance.