Advanced recommendations for Rightsizing

Overview

Rightsizing aligns your provisioned Azure resources with actual workload requirements, so you get optimal cost efficiency without sacrificing performance. Advanced recommendations give you personalized, configurable rightsizing suggestions from Cost Analyzer, going beyond the Azure Advisor-based Basic recommendations view. Navigate to Cost group > Optimization > Rightsizing > Advanced recommendations to get started.

Business value

Advanced recommendations reduce unnecessary cloud expenditure by flagging oversized and underutilized resources with configurable, resource-specific analysis, and let you act on those recommendations: apply, schedule, notify a team, or generate a ready-to-run script, without leaving Cost Analyzer.

Prerequisites

Required permissions

Applying a recommended SKU additionally requires Contributor access on the Service Principal, regardless of role. Roles are created under Cost Analyzer > User management > Roles > Add role, and access can be scoped to specific cost management groups.

How it works

Cost Analyzer analyzes actual resource utilization against the provisioned SKU and generates a rightsizing recommendation for each compatible resource. You can customize the metrics and thresholds that drive those recommendations, and act on them directly from within Cost Analyzer. The following resource types support Advanced recommendations:

• App service plan
• App Service Plan (Workflow Standard SKU)
• Azure Database for MySQL flexible server
• Cosmos SQL Container
• Cosmos SQL Database
• Cosmos Table
• Log Analytics workspace
• SQL Database
• SQL Elastic pool
• SQL Managed Instance
• Virtual Machine
• Virtual Machine scale set

Cost Analyzer produces four recommendation types:

1. Upgrade: the resource frequently hits the maximum capacity of its current SKU. Moving to a higher SKU improves performance and prevents resource contention.
2. Downgrade: the resource consistently operates well below its capacity. Moving to a lower SKU reduces cost without impacting performance.
3. Idle: the resource has seen no meaningful usage for an extended period. It can be deleted if it is no longer required.
4. No change: the resource utilization aligns well with the current SKU. No action is needed.

The Next recommendation at timestamp shown in the header indicates when Cost Analyzer will next evaluate resources and refresh recommendations for the group.

Log Analytics workspace recommendations

Log Analytics workspace recommendations work differently from other resource types. Instead of a SKU resize, Cost Analyzer evaluates whether switching from Pay-as-you-go (PerGB2018) to a Commitment Tier would reduce cost based on the workspace's ingestion volume over the last 30 days.

When a workspace's average daily ingestion exceeds 100 GB/day, Cost Analyzer recommends a Commitment Tier. Azure Log Analytics offers the following Commitment Tiers, each offering a fixed discount over Pay-as-you-go:

Cost Analyzer recommends the lowest Commitment Tier that covers the workspace's observed ingestion volume. Applying the recommendation switches the workspace's pricing model in Azure, and the discount applies to all ingestion above the committed threshold.

Rightsizing settings thresholds do not apply to Log Analytics workspaces. The recommendation is based solely on ingestion volume against the 100 GB/day threshold.

Understanding resource usage charts

Click any resource listed under Advanced recommendations to view its usage patterns. Use the Time settings option to adjust the time range and focus on a specific interval. The area chart shows both the current and recommended SKUs and lets you navigate between intervals for detailed comparison. Data points that exceed or fall below the thresholds configured in Rightsizing settings are visually distinguished:

• Yellow: the data point surpassed the upgrade threshold, indicating the resource is under pressure and a higher SKU is recommended.
• Blue: the data point fell below the downgrade threshold, indicating the resource is underutilized and a lower SKU is recommended.

Resource status calendar

Clicking an Advanced recommendation also opens a resource usage calendar view that shows the status of the resource over time, with multiple toggle modes available for accessibility. Resource status is one of:

• Active
• Idle

The Agenda view displays a day's active and idle hours in a grid format for a granular look at utilization patterns.

Status is determined by the following metric for each resource type:

• App service plan: CPU Percentage
• Azure Database for MySQL flexible server: Host CPU percent
• Cosmos SQL Container: Normalized RU Consumption
• Cosmos SQL Database: Normalized RU Consumption
• Cosmos Table: Normalized RU Consumption
• Disk: Disk Read Write
• Log Analytics workspace: Billable ingestion (GB/day)
• SQL Database: DTU consumption
• SQL Elastic Pool: DTU consumption
• SQL Managed Instance: CPU Percentage
• Virtual Machine: CPU Percentage
• Virtual Machine Scale set: CPU Percentage

Idle recommendation resource types

The following resource types, when detached from all Azure services, support only Idle recommendations, letting you identify and remove abandoned resources accruing cost without serving any workload:

1. Disks
2. Public IP Addresses
3. Virtual Network
4. Subnet

A virtual network without any subnet is considered for idle recommendation.

Cost intelligence view

The Cost intelligence view structures resource data into distinct categories and visual sections, giving you a detailed overview of cost distribution and utilization. It highlights key optimization opportunities by identifying areas of high or low resource efficiency, so you can monitor spending patterns, evaluate cost performance, and take data-driven optimization actions. See Cost analysis to learn more about the Cost intelligence view's capabilities. To access this view, navigate to Optimization > Rightsizing > Advanced recommendations within a cost management group, then click the View resource details icon next to a recommended resource.

Explain with AI

Each Advanced recommendation includes an Explain with AI option. Hover over a recommendation and click the icon to open a detailed AI-generated analysis panel.

The panel covers seven areas:

• Savings summary: recommendation type (Downgrade, Upgrade, Idle, or No change), current monthly cost, monthly and annual savings, and a context note explaining the basis of the recommendation.
• Metric analysis: the monitored metrics used to generate the recommendation (such as CPU percentage and DTU percentage), with average, peak, downgrade threshold, and upgrade threshold values, each labeled against its threshold status.
• Recommendation summary: what SKU change is being recommended, why the workload qualifies based on monitored signals, and the financial impact of keeping the current tier versus acting.
• Implementation steps: step-by-step actions with an estimated time and the tool required (Turbo360 Cost Analyzer, Turbo360 Rightsizing Settings, or Azure Monitor).
• Risk and impact assessment: Performance Risk, Availability Risk, and an Overall Safety rating, with guidance on what to monitor after the change.
• Rollback plan: how to revert the change, including early warning signals and confirmation that data, logins, and configuration are preserved.
• Recommendation justification: why the recommendation was generated based on monitored metric samples, and the recommended timing to minimize operational impact.

Steps

Navigate to Cost group > Optimization > Rightsizing > Advanced recommendations to get started. Actions on individual recommendations are available from the Actions dropdown on each row. Page-level controls are available from the ⋯ menu at the top right of the recommendations list.

Understanding the recommendations list

The recommendations list includes the following columns to help you assess and prioritize action:

• Identified on: the date when the current recommendation type was first detected for this resource during a scheduled refresh. This date updates only when the recommendation type changes, such as when a resource transitions from No change to Upgrade, Downgrade, or Idle. If the recommendation type has not changed between refreshes, this date stays fixed regardless of how many cycles have run. Recommendations refresh every 48 hours; there is no recommendation history.
• Age: how long the recommendation has been open since it was first identified, color-coded by urgency: green (up to 30 days), orange (31 to 90 days), and red (over 90 days, stale).

Schedule action

Scheduling an action resizes selected resources to their recommended SKU at a specified date and time, with optional notification channels to alert your team when the action runs.

1. Select one or more resources using the row checkboxes.
2. Click Actions > Schedule action.
3. In the Add schedule action dialog, set the Schedule time. Select a date and time for the resize to execute.
4. Optionally, select Delete idle resources to automatically delete any resources with an Idle recommendation when the scheduled action runs.
5. Review the Resource details table, which lists each selected resource with its Name, Recommendation type, Resource group, Subscription, Current SKU, Recommended SKU, and an optional Comments field.
6. Under Notification settings, select one or more channels (Teams, ServiceNow, Slack, or others) to receive an alert when the action completes.
7. Click Save.

Saved scheduled actions can be reviewed and managed from ⋯ > Scheduled actions.

Ignore a recommendation

Ignoring a recommendation removes it from the active list when it is not relevant to your environment. Unlike Dismiss in Basic recommendations, ignored recommendations are tracked separately and can be reviewed at any time.

1. Select the resource you want to ignore using the row checkbox.
2. Click Actions > Ignore.

Recommendation settings

Recommendation settings lets you control how Advanced recommendations are generated and applied for the current cost management group. To open it, click the Recommendation settings button in the top-right corner of the Advanced recommendations view.

After making changes, click Update to save. The maximum evaluation duration that can be set is 90 days.

FAQs

1. What is the difference between Advanced recommendations and Basic recommendations for Rightsizing?
   Basic recommendations surface Azure Advisor's cost recommendations as-is for your subscriptions and resource groups, with no configuration required. Advanced recommendations are generated by Cost Analyzer's own utilization analysis, using configurable metrics and thresholds you define in Rightsizing settings. Advanced recommendations support direct actions (schedule, script, notify) and cover a broader set of resource types including SKU-level analysis. Use Basic for a quick Advisor-based view; use Advanced for configurable, actionable rightsizing analysis.

2. Does scheduling an action apply the SKU change immediately?
   No. Scheduling an action sets a future date and time for the resize to execute. The resource continues running at its current SKU until the scheduled time. If you need to apply a change immediately, use the Az CLI script or make the change directly in the Azure portal.

3. What is the difference between Ignore and Dismiss?
   Ignore is the Advanced recommendations equivalent of Dismiss in Basic recommendations. In Advanced recommendations, ignored recommendations are tracked under View ignored recommendations and can be restored. The underlying mechanism is specific to Cost Analyzer's own analysis and does not propagate to Azure Advisor, since Advanced recommendations are generated by Cost Analyzer, not Azure Advisor.

4. How does the Log Analytics workspace recommendation work?
   When a workspace's average daily ingestion over the last 30 days exceeds 100 GB/day, Cost Analyzer recommends switching from Pay-as-you-go (PerGB2018) to a Commitment Tier. Commitment Tiers offer a 25-35% discount over Pay-as-you-go. Cost Analyzer selects the lowest tier that covers the observed ingestion volume. Rightsizing settings thresholds do not apply to this resource type.

5. Can I configure the ingestion threshold for Log Analytics recommendations?
   No. The 100 GB/day ingestion threshold for Log Analytics workspace recommendations is fixed and cannot be adjusted in Rightsizing settings.

Related articles

• Basic recommendations for Rightsizing
• AI Agents in Cost Analyzer
• Advanced recommendations for Reservations
• OpenAI setup in Turbo360