Leading public cloud providers offer several unique cloud services—each with different commitment and discounting options. With so many choices, it’s difficult to know what options are best for your business. To break it down, we’ve compared cloud discounts, commitments, and reservations between the three primary cloud providers—Azure, GCP, and AWS.
What is a reservation?
A Reservation, Savings Plan, or Committed Use Discount is a resource commitment with a fixed term in exchange for discounted pricing. Purchasers benefit because they can pay less money, and cloud providers benefit because they can plan usage capacity. However, savings opportunities vary depending on the cloud provider.
Azure, Google Cloud, and AWS use different names for reservations. Azure Reservations can apply to compute, storage, or app services. GCP Committed Use Discounts (CUDs) apply to compute services or Cloud SQL, while flat-rate purchases apply to BigQuery (BigQuery Reservations). Finally, AWS offers Reservations for EC2, RDS, ElastiCache, Elasticsearch, Redshift, and DynamoDB, and Savings Plans for EC2, Fargate, and Lambda, where you’re committing to a minimum dollar spend per hour. See our blog for an in-depth comparison of AWS Reservations and Savings Plans.
What is reserved capacity?
Reserved capacity is a guarantee that you’ll be able to spin up your requested infrastructure in a specific location in the unlikely event that there are capacity constraints. Depending on the cloud provider, reserved capacity can be coupled or decoupled with reservations.
- Azure: Virtual Machine reservations are given prioritized compute capacity, whereas the reservations for database services (e.g. Cosmos DB) are referred to as reserved capacity. If you purchase an Azure Reservation, you’ll automatically be able to launch your infrastructure with that specific configuration.
- GCP: Reserved capacity is decoupled from reservations. When you purchase Committed Use Discounts, it’s not guaranteed you can launch an instance in the specified region or configuration. If you’d like that guarantee, you’ll need to purchase a zonal reservation. It’s important to note that zonal reservations don’t have a contract term. You can purchase one in the morning and terminate it in the evening. This is an opportunity to purchase reservations and reserved capacity separately to operate on an as-needed basis.
- AWS: Some types of AWS Reservations come with capacity benefits (Zonal Reserved Instances for example), but Savings Plans and Regional Reserved Instances don’t have this benefit. AWS also offers the ability to purchase only the capacity guarantee—called a Capacity Reservation. Just like Google’s zonal reservations, you can end them whenever you like.
Given this overview, we’ll provide more detail into the different discount, commitment, and reservation options for each cloud provider.
Discounts, commitments, and reservations with Azure
When you purchase Azure Reservations, you’ll receive a discounted price in return for committing to pay for those resources for one to three years. You can purchase more than 15 different services with Azure reservations. Among CloudHealth customers, the most common purchases are Reserved Virtual Machine Instances.
Azure offers three different discount models for reservations based on scope:
- Single Resource Group Scope: The reservation you purchase will only apply to usage in the specified resource group. If the reservation isn’t being used, it won’t apply to usage outside of that.
- Single Subscription Scope: The reservation you purchase lives within a subscription and applies to all usage within that subscription. If it isn’t used, it won’t apply to any other subscriptions.
- Shared Scope: The reservation discount applies to all usage within a subscription or outside of it to matching resources.
It’s important to note Azure’s flexibility with reservations and billing. With Azure, you can update your scope during the life of a reservation, exchange or cancel reservations up to $50k a year, and receive discounts for all upfront or monthly payment plans.
Discounts, commitments, and reservations with Google Cloud Platform
With GCP Committed Use Discounts (CUDs), when you purchase compute resources (vCPUs, memory, GPUs, and local SSDs), you receive a discounted price in return for committing to pay for those resources for one to three years.
Google offers two different models for CUDs based on scope:
- Billing Account Scope: Applies to usage anywhere in the Billing Account, regardless of which project purchased it.
- Project Scope: Applies only to usage within the project that purchased it.
Once committed to a contract with GCP, you’re billed each month for the duration of the term whether you use the services or not. When you make a commitment to an instance family, you’re committed to that instance family for the life of the commitment. Unlike Azure, you can’t make changes or cancellations.
Depending on the service, GCP offers different payment plan opportunities. With compute, there’s only No Upfront—you can’t purchase Partial or All Upfront CUDs. With BigQuery Reservations, it’s flat-rate pricing based on usage, and with CloudSQL, it’s an hourly commitment model, which is a recent update since June 2020 and is similar to AWS’ Savings Plan model. You can learn more about the differences in this article comparing CUDs and AWS Reserved Instances.
Google also offers discounts for running an instance for a significant portion of the billing month with Sustained Use Discounts (these don’t technically qualify as reservations because they don’t have an end date). However, you can’t combine Sustained Use Discounts and CUDs. Google applies all your CUDs first, and will then apply Sustained Use Discounts if remaining usage is available. For more information, see Google’s complete pricing breakdown of Sustained Use Discounts.
Discounts, commitments, and reservations with AWS
AWS Reserved Instances (RIs) provide savings on compute costs for either one or three-year commitments. Concerning scope, reservations can be broken down by Region or Availability Zone:
- Regional Scope: Discounts apply to instance usage in any Availability Zone in the specified region, and to any instance within the same family, regardless of size.
- Availability Zone (AZ) Scope: Discounts only apply to usage within a specific Availability Zone, and there’s no flexibility on the instance size or type.
As we mentioned above, capacity reservations are coupled with AZ-scoped RIs but not with regionally scoped RIs.
Reserved Instances can be either Standard or Convertible. Standard RIs can be resold on the AWS RI Marketplace, but they can’t be exchanged with a different instance family type. Convertible RIs cannot be resold, but they can be exchanged across instance family, instance type, platform, scope, or tenancy.
AWS supports All Upfront, Partial, and No Upfront reservation payments. Depending on the plan and how much you pay, you receive different discounts. Generally, the more you pay upfront, the more significant discount you’ll receive.
As you can tell from all the permutations of Reserved Instances (more than 8,000!), AWS customers wanted a simpler, more flexible way to receive discounts for their AWS usage. So in November 2019, AWS changed the game with AWS Savings Plans. Savings Plans are also based on a one or three-year commitment, but rather than a utilization commitment, it’s a monetary commitment based on how much you spend per hour.
Currently, Savings Plans apply to EC2, Fargate, and Lambda services, though Savings Plans for other services are expected to be available in the future. While we have seen a decrease in RI purchases since the introduction of Savings Plans, some AWS customers are still using RIs for a number of reasons. This article highlights feedback from AWS customers who are still using RIs.
Customers can choose how much they wish to commit to (minimum $0.001 per hour per year) and layer Savings Plans on top of one another. For example, you can purchase a $10 per hour Savings Plan and as consumption increases or Reserved Instances expire, purchase further Savings Plans to maximize coverage and minimize costs.
Take advantage of discounts, commitments, and reservations
The key to taking advantage of discounts and reservations effectively is to use the most appropriate deal wherever available, which often means purchasing reservations from multiple cloud providers (if you’re running in multiple clouds). But to know which reservation you should choose to achieve maximum cost savings, you need to have total visibility into your multicloud environment and insight into how your instances are being used.
In order to do this, you need an agnostic multicloud management solution that not only enables you to analyze the performance and utilization of your entire multicloud infrastructure, but also provides tailored recommendations and automation capabilities.
To learn more about managing costs across your multicloud environment, see our eBook: 9 Best Practices for Reducing Spend in Your Multicloud Environment