Hello, I am just trying to understand how does Azure generate reservation (RI) recommendations ? Does Azure takes into account rightsizing recommendations in Azure Advisor on resources for which it is generating reservation (RI) recommendations? How many days of resources usage data does Azure analyze at to generate reservation (RI) recommendations? Are reservation (RI) recommendations generated on demand when you click on add to cart in the Azure Reservation portal ? I feel Azure just look at only the sizes/tiers of resources but not their actual utilization or rightsizing needs while generating reservation (RI) recommendations? How can we add value to Azure reservation (RI) recommendations or how can we make our RI purchases more efficient ? Appreciate your response. Thank you.

Azure generates reservation (RI) recommendations by evaluating the hourly usage of your resources over the past 7, 30, and 60 days (can be adjusted) . Based on this usage data, it simulates your costs with and without reservations for different quantities and recommends the quantity that maximizes the savings. Azure does take into account rightsizing recommendations in Azure Advisor on resources for which it is generating reservation (RI) recommendations. The recommendations account for existing reservations and savings plans, so previously purchased reservations and savings plans are excluded when providing new recommendations. Reservation (RI) recommendations are not generated on demand when you click on add to cart in the Azure Reservation portal, they are pre-calculated based on your past usage data and presented to you in the portal. While it’s true that Azure looks at the sizes/tiers of resources, it also considers their actual utilization. If your resources are shut down regularly, the simulation can’t find any savings, and no purchase recommendation is provided. So, it’s not just about the size/tier of the resource, but also about how consistently and efficiently it’s being used. For more details please refer: https://learn.microsoft.com/en-us/azure/cost-management-billing/reservations/determine-reservation-purchase . https://learn.microsoft.com/en-us/azure/cost-management-billing/reservations/reserved-instance-purchase-recommendations . https://azure.microsoft.com/en-us/updates/azure-advisor-vmvmss-rightsizing-recommendations-with-custom-lookback-period/ . To add value to Azure reservation (RI) recommendations and make your RI purchases more efficient, consider the following: Analyse your usage data : Understand your workload and usage patterns. This will help you select the right size and quantity for your reservations. Consider Azure Reserved Instances : If you have predictable and steady workloads, consider reserving your resources for one or three years. Reserved Instances provide capacity reservation, ensuring availability even during peak usage, and reduce costs compared to on-demand pricing. Perform Performance Testing and Benchmarking : Run your workload on different resource sizes and measure metrics such as response time, throughput, and resource utilization. Use Azure Advisor : It provides personalized recommendations based on your usage and configurations to help you optimize your Azure deployments. Avoid underutilization : Purchasing more capacity than your historical usage results in an underutilized reservation. You should avoid underutilization whenever possible. Monitor and adjust : Regularly review your reservation utilization and adjust your reservations as needed. Some Additional helpful documentations: https://learn.microsoft.com/en-us/azure/cost-management-billing/reservations/determine-reservation-purchase . https://learn.microsoft.com/en-us/azure/cost-management-billing/reservations/reservation-utilization . https://learn.microsoft.com/en-us/azure/advisor/view-recommendations

Logic behind Azure reservation (RI) recommendations

Accepted answer

Walmiki Nirmalkar 80 Reputation points Microsoft Employee

2024-06-22T08:38:03.3266667+00:00
Azure generates reservation (RI) recommendations by evaluating the hourly usage of your resources over the past 7, 30, and 60 days (can be adjusted). Based on this usage data, it simulates your costs with and without reservations for different quantities and recommends the quantity that maximizes the savings.

Azure does take into account rightsizing recommendations in Azure Advisor on resources for which it is generating reservation (RI) recommendations. The recommendations account for existing reservations and savings plans, so previously purchased reservations and savings plans are excluded when providing new recommendations.

Reservation (RI) recommendations are not generated on demand when you click on add to cart in the Azure Reservation portal, they are pre-calculated based on your past usage data and presented to you in the portal. While it’s true that Azure looks at the sizes/tiers of resources, it also considers their actual utilization.

If your resources are shut down regularly, the simulation can’t find any savings, and no purchase recommendation is provided. So, it’s not just about the size/tier of the resource, but also about how consistently and efficiently it’s being used.

For more details please refer:

https://learn.microsoft.com/en-us/azure/cost-management-billing/reservations/determine-reservation-purchase.

https://learn.microsoft.com/en-us/azure/cost-management-billing/reservations/reserved-instance-purchase-recommendations.

https://azure.microsoft.com/en-us/updates/azure-advisor-vmvmss-rightsizing-recommendations-with-custom-lookback-period/.

To add value to Azure reservation (RI) recommendations and make your RI purchases more efficient, consider the following:

Analyse your usage data: Understand your workload and usage patterns. This will help you select the right size and quantity for your reservations.

Consider Azure Reserved Instances: If you have predictable and steady workloads, consider reserving your resources for one or three years. Reserved Instances provide capacity reservation, ensuring availability even during peak usage, and reduce costs compared to on-demand pricing.

Perform Performance Testing and Benchmarking: Run your workload on different resource sizes and measure metrics such as response time, throughput, and resource utilization.

Use Azure Advisor: It provides personalized recommendations based on your usage and configurations to help you optimize your Azure deployments.

Avoid underutilization: Purchasing more capacity than your historical usage results in an underutilized reservation. You should avoid underutilization whenever possible.

Monitor and adjust: Regularly review your reservation utilization and adjust your reservations as needed.

Some Additional helpful documentations:

https://learn.microsoft.com/en-us/azure/cost-management-billing/reservations/determine-reservation-purchase.

https://learn.microsoft.com/en-us/azure/cost-management-billing/reservations/reservation-utilization.

https://learn.microsoft.com/en-us/azure/advisor/view-recommendations
Please sign in to rate this answer.
Anil Kumar 265 Reputation points

2024-06-24T07:35:06.6766667+00:00

Thank you @Walmiki Nirmalkar for taking time to help and clarify.

Glad to mark your response as an answer !

Request you to consider answering below follow up questions as well

"Azure generates (RI) recommendations ...7, 30, and 60 days (can be adjusted)."

How Can I adjust it to look for only 60 days usage data?

Does it get usage data from Azure Metrics database which stores data for 90 days at no cost?

Can I make it 90 days as my some systems are used at the end of quarter?

"Reserved Instances provide capacity reservation, ensuring availability even during peak usage" - I believe reservation is just a billing mechanism and no way related to capacity. May be for that reason, capacity reservation is different service is Azure. Am i missing something here?

Another question is can I make Azure Advisor to generate RI recommendation's at the resource group level ? Currently its at Subscription level.

Looking forward for your response.

#StayBlessednHappy

Walmiki Nirmalkar 80 Reputation points Microsoft Employee

2024-06-24T09:18:52.9233333+00:00

Sure, I’d be happy to answer your questions:

<1> How Can I adjust it to look for only 60 days usage data?

You can modify the lookback period for Azure Advisor recommendations by following the instructions shown in the screenshot below.

Changing lookback period on RI purchasing blade:

Reference: Azure reservation recommendations - Microsoft Cost Management | Microsoft Learn

Changing the lookback period in azure advisor recommendation blade:

Reference: Cost recommendations - Azure Advisor | Microsoft Learn

Only VM/VMSS rightsizing recommendations look back period can be changed to 90 days:

Reference: Azure Advisor VM/VMSS right-sizing recommendations with custom lookback period | Azure updates | Microsoft Azure

<2> Does it get usage data from Azure Metrics database which stores data for 90 days at no cost?

Azure Advisor utilizes your cost and billing data from the past 7, 30, and 60 days to provide cost-saving recommendations, such as Reserved Instance (RI) suggestions.

However, recommendations for resizing VMs/VMSS are based on Azure metrics database which stores data for 90 days at no cost, as they need to analyze metrics like CPU and memory usage percentages to recommend you the right size of VM/VMSS.

Reference: Determine what Azure reservation you should purchase - Microsoft Cost Management | Microsoft Learn

<3> Can I make it 90 days as some systems are used at the end of quarter?

You can change the lookback period to 60 days for RI and other cost-saving recommendations. However, for VM/VMSS resizing recommendations, you can extend the lookback period to 90 days, as shown in the screenshot above.

Resource: Azure reservation recommendations - Microsoft Cost Management | Microsoft Learn and Azure Advisor VM/VMSS right-sizing recommendations with custom lookback period | Azure updates | Microsoft Azure

<4> Reserved Instances vs Capacity Reservation: You’re correct. Azure Reserved Instances (RIs) provide a discounted price for a 1-year or 3-year commitment and are not tied to specific instances. On the other hand, Azure’s On-demand Capacity Reservation allows you to reserve compute capacity for any duration without a long-term commitment. So, while RIs provide cost benefits, Capacity Reservations ensure resource availability.

Reference:

https://learn.microsoft.com/en-us/azure/cost-management-billing/reservations/save-compute-costs-reservations

https://learn.microsoft.com/en-us/azure/virtual-machines/capacity-reservation-overview

<5> Azure Advisor RI Recommendations at Resource Group Level: Azure Advisor allows you to select Subscriptions and Resource Groups that matter to you and your organization. You only see recommendations for the subscriptions and resource groups that you select. So, yes, you can configure Azure Advisor to generate RI recommendations at the resource group level.

Reference: https://learn.microsoft.com/en-us/azure/advisor/view-recommendations

You can generate the recommendation on any scope of your choice such as Subscription or resource group.

Go to your subscription >> Advisor recommendations:

Go to your resource group >> Advisor recommendations:

When purchasing a reservation, the reservation purchase blade offers recommendations based on the scope you selected (Shared (entire billing account/profile), Management Group, single subscription, or resource group). You can adjust the lookback period to your preference (7, 30, or 60 days). The recommendations will appear under the recommended view with suggested quantities, but you can bypass these recommendations and purchase any RI of your choice from all products.

I hope this helps!
Sign in to comment

Share via

Logic behind Azure reservation (RI) recommendations

0 additional answers