The cloud now has more resources than ever, and from it, the development of cloud flexibility. From these developments comes the progress of enabling reservations on compute, storage, databases, and other services that make up overall consumption as the cloud evolves. These commitments to resources can end up saving your organization in up to 40% on costs compared to on demand.
But as the breadth of reservations continues to expand, tracking, monitoring, and performing cost optimization becomes much harder, up to the point that it may take a Managed Service Provider (MSP) to track your services, quantity, sizes & expiration dates for you.
Commitments are just that. You must utilize those committed resources to receive any of the benefit for at least 1 year, and in many cases, this may not be the best thing for your infrastructure and its needs. Luckily, there are other options to help with this inconvenience.
The most flexible model of commitment in Azure is called, Azure Savings Plans. Azure Savings Plans allows you to forgo several of the headaches that are highlighted above and focus on other needs in your organization that are likely to need more time and attention.
The Savings Plan is a concept that was invented by Amazon Web Services and was an enormous success in allowing technical and financial teams reduce the stress of tracking and managing the commitments. In October of 2022, Microsoft introduced the Azure Savings Plan, its own version of savings plan for Microsoft Azure compute services such as: Virtual Machines, Azure App Service, Azure Functions premium plan, Azure Container Instances, and Azure Dedicated Host.
The Azure Savings Plans is a flexible pricing model for compute services that can reduce costs by up to 65% on select services vs. the on-demand pricing of the normal Pay-As-You-Go model. Azure Savings Plans allow the teams to commit to a fixed hourly spend on compute services for one or three-year terms. This eliminates the headache of spending time on allocation behind the scenes and allocates the savings to the appropriate resources usage for you.
How it Works
The user commits to spend an amount every hour of the day, let us say $10. Azure calculates the cost of each individual VMs (Virtual Machine) hourly rate and applies that cost against the Savings Plan, once the committed amount is consumed by the existing VMs the remaining cost is charged with Pay-As-You-Go prices. You essentially get lower prices by agreeing to spend a fixed hourly commitment on compute services for either one or three years.
If your usage is less than or equal to your hourly commitment, you achieve lower saving plan prices. If your usage is less, you are still committed to paying your full hourly commitment. However, no additional fees are charged, and your usage is fully covered,
If your usage is greater than the hourly commitment, the usage up to your hourly commitment is billed at lower, savings plan prices and included in the cost of your savings plan. All of the additional usage is charged at pay-as-you-go rates and invoiced separately.
The savings plan coverage is applied to the more expensive resources first and then to the smaller units to ensure it provides the highest savings possible. If you happen to have purchase multiple savings plans, the savings plan is first applied to the hourly usage from the service with the greatest savings percentage. If more than one of your plans has the same savings percentage, the plan with the narrowest scope is applied first. From narrowest to broadest, a plan can be applied to resource group, single subscription, management group, or shared tenant. Once the benefit from the commitment has been fully applied, if more usage from the service remains, the broader-scoped savings plan is then applied. When this savings plans benefit has fully covered that service's usage, the process continues to cover usage from the service with the next highest savings percentage.
If in an hour the commitment cost is not reached – that is wasted cost.
Azure has now adopted this model into its cloud commitment offering, but with one distinct improvement, for a limited time, they allow converting existing reservations into the new Savings Plan.
The way this works is as follows.
The remaining commitment is calculated to an hourly rate and that becomes the new hourly commitment of rate, which will remain until the original expiration date of the Reservation [ex: a 4 D2s_v4 reservation is $192.34 per month, divide that by 730 hours (monthly hours) and the hourly commitment is $0.26.
While the offer to convert reservations to Savings Plans is in effect, Microsoft is also not charging the cancellation fee for RI (Reserved Instances) cancellations.
Now all this sounds like a wonderful product, and you may ask yourself, “Why is Azure doing this and what is the catch?” Well, there are two answers to that question:
- Azure Savings Plans offer flexibility and an easier way to achieve savings on compute consumption without the need to worry if you want to move from one virtual machine type to another (Dsv4 to Dsv5, or from D4s_v4 to E8_v3), since the saving plan will cover the cost of both based on the hourly spending.
- Please note that the discount percentage provided by Savings Plans can be lower than that of regular reserved instances. Although Azure states you can save up to 65% on pay-as-you-go prices, savings rates range from as high as 65% but can go as low as 11%. Some reserved instances can provide a 41% discount (depending on the location and virtual machine type), a Savings Plan discount for the same machine and the same location is 20%, which is a significant reduction. Also, when converting reservations to a savings plan, the resulting hourly commitment might not be as large of a benefit to cover the virtual machines that were utilized in prior reservations.
This offer for exchanges of reserve instances for compute services is ending on January 1, 2024. So be sure to take advantage of this flexible pricing plan if it sounds beneficial or advantageous for your team, infrastructure, and overall cloud management.
About the author: Joe Thomas is a FinOps Engineer at 2bcloud, a born-to-the-cloud, next-gen managed service provider (MSP) that works with fast growing, cloud-native startups. 2bcloud leverages its multi-cloud expertise working with Amazon Web Services (AWS) and Microsoft Azure to help customers grow revenue, increase efficiency, reduce cost, and deliver scalable experiences.
Edited by
Erik Linask
