Optimizing Azure Costs with Azure Advisor

Azure Advisor is a powerful tool that can help you identify and address opportunities to optimize your Azure infrastructure and reduce your overall cloud spend. By analyzing your resource utilization and making targeted recommendations, Advisor empowers you to make informed decisions that can lead to significant cost savings.

Identifying Underutilized Virtual Machines and Scale Sets

One of Advisor’s key capabilities is its ability to detect underutilized virtual machines (VMs) and virtual machine scale sets (VMSS). Using machine learning algorithms, Advisor examines metrics like CPU, memory, and network utilization to identify resources that are not being efficiently used. It then provides recommendations to either shut down these idle resources or resize them to a more appropriate and cost-effective SKU.

Shutdown Recommendations

Advisor will recommend shutting down VMs or VMSS instances that have been completely idle over the past seven days. The criteria for these recommendations include:

  • P95 of the maximum value of CPU utilization across all cores is less than 3%
  • P100 of average CPU utilization over the last 3 days is 2% or less
  • Outbound network utilization is less than 2% over a seven-day period

By shutting down these idle resources, you can realize immediate cost savings.

Resize Recommendations

In cases where your workloads could be supported by a less expensive VM or VMSS SKU, Advisor will provide recommendations to resize the resource. Advisor evaluates CPU, memory, and network utilization to determine the optimal SKU that meets the performance requirements of your workloads without over-provisioning. Key factors in these recommendations include:

  • For user-facing workloads, keeping CPU and outbound network utilization below 40% and memory utilization below 60% on the new SKU.
  • For non-user-facing workloads, keeping CPU, network, and memory utilization below 80% on the new SKU.
  • Ensuring the new SKU has the same capabilities (e.g., accelerated networking, premium storage) as the current one.
  • Prioritizing instance count changes over SKU changes for VMSS resources, as they are more easily actionable.

Burstable Recommendations

Advisor also evaluates whether your workloads could benefit from the use of burstable VM SKUs, which provide a baseline level of CPU performance and the ability to burst above that baseline when needed. These burstable SKUs are often more cost-effective than general-purpose SKUs, especially for workloads with low average utilization but occasional spikes.

Considerations and Limitations

While Advisor’s recommendations can provide significant cost savings, there are a few important factors to keep in mind:

  • The estimated savings are based on retail rates and may not reflect any discounts or pricing structures specific to your Azure account.
  • The recommendations do not take into account any existing Reserved Instance or Savings Plan commitments you may have, which could impact the actual cost implications.
  • In some cases, the recommended actions may not be feasible or appropriate, such as when a VM has been provisioned to handle future increases in demand or is used for testing and development purposes.

To address these limitations, Advisor allows you to configure VM and VMSS recommendations by subscription, enabling you to adjust the CPU utilization thresholds used to generate the recommendations.

Conclusion

Azure Advisor is a valuable tool for optimizing your Azure infrastructure and reducing your cloud costs. By leveraging Advisor’s machine learning-powered recommendations, you can identify and address underutilized resources, resize VMs and VMSS instances to more appropriate and cost-effective SKUs, and take advantage of burstable VM options. While the recommendations should be evaluated in the context of your specific requirements and existing commitments, Advisor can be a powerful ally in your efforts to maximize the efficiency and cost-effectiveness of your Azure deployments.

For more information, be sure to check out the Advisor cost recommendations and Introduction to Advisor documentation.

Source: Reduce service costs using Azure Advisor