Cloud Advisor Categories and Recommendations

Learn about each Cloud Advisor category and its recommendations.

Following are Cloud Advisor categories:

Cost Management Recommendations

Cost management recommendations help you identify and alter underutilized resources so that you can save money without degrading performance. This section lists each recommendation that comes with a cost savings estimate and describes how Cloud Advisor calculates that estimate.

How Cost Savings Estimates Are Calculated

Cloud Advisor estimates cost savings for applicable recommendations. Each cost savings estimate indicates how much lower your costs become when you implement the recommendation.

Cloud Advisor also provides a percentage cost savings of the expected monthly bill for the tenancy. The expected bill is based on data from current usage and the previous month. If less than a full month of data is available, then Cloud Advisor pro-rates cost savings using data starting on the first day of the current month. New resources have less than a full month of data when they are created in the current month or later in the previous billing month than needed for a full month of data.

Change compute instances to burstable

The Change compute instances to burstable recommendation indicates bursty traffic in single-OCPU E3 Flex compute instances. Bursty traffic occurs when instances have fairly low usage with only occasional spikes of usage. Adjusting instances to the suggested burstable configuration saves you money without degrading performance.

Recommendation generation logic

Cloud Advisor detects bursty single-OCPU E3 Flex instances using CPU utilization metrics (Max and P95 statistics). In a 24-hour period, Cloud Advisor reviews 48 data points for CPU utilization per instance, generating this recommendation when only a few data points (5 or fewer) have Max CPU utilization greater than a threshold (50% or 12.5%). The P95 CPU utilization value in this duration should be less than 40% (for 50% threshold) and 10% (for 12.5% threshold) to ensure that the usage is fairly low (non-spiky) for most part of the duration. This recommendation suggests adjusting the instance to burstable at the percentage indicated by the exceeded threshold. Burstable 50% is suggested for the 50% threshold; burstable 12.5% is suggested for the 12.5% threshold.

To customize the logic for this recommendation, see To edit a global recommendation override.

Savings calculation

Cloud Advisor calculates cost savings for an instance as the difference between the suggested burstable configuration and the standard configuration. For example, if the suggested burstable configuration is 12.5%, then the new cost would be 12.5% of the standard configuration.

How to implement

Do one of the following:

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow or manual steps to change the instance to a burstable configuration), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).
Delete idle compute instances

The Delete idle compute instances recommendation indicates that some compute instances are unused. Deleting unused compute instances saves you from paying for instances that you aren't using. Boot volumes are retained.

Recommendation generation logic

The past 7 days of data is used. The logic depends on the shape.

For supported VM shapes:

  • CPU Utilization P95 over all days is < 3%
  • Memory utilization max over all days is < 8%
  • Network utilization max over all days is < 3%

For supported BM shapes:

  • CPU Utilization P95 < 2%
  • Memory utilization max < 5%
  • Network utilization max < 3%
Supported shapes

Following are the shapes supported for Cloud Advisor scans using Delete idle compute instances:

VM shapes:

  • VM.Standard.B1.1
  • VM.Standard.E2.1
  • VM.Standard1.1
  • VM.Standard2.1
  • VM.Standard.E3.Flex
  • VM.Standard.E4.Flex
  • VM.Optimized3.Flex

BM shapes:

  • BM.Standard1.36
  • BM.Standard2.52
  • BM.Standard.B1.44
  • BM.Standard.E2.64
  • BM.Standard.E3.128
  • BM.Standard.E4.128
  • BM.DenseIO1.36
  • BM.DenseIO2.52

Savings calculation

For each compute instance, Cloud Advisor multiplies the billed usage by the unit price.

How to implement

Do one of the following:

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow or manual deletion), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).
Delete unattached block volumes

The Delete unattached block volumes recommendation indicates that unattached block volumes exist in your tenancy. Attaching or deleting unattached block volumes reduces costs.

Recommendation generation logic

Cloud Advisor scans the tenancy once per day for block volumes that are not attached to any compute resource. If the volume is unattached for fourteen consecutive days, Cloud Advisor recommends that you delete it.

Savings calculation

Cloud Advisor estimates how much money would be saved by deleting the unattached block volume. For the resource that the recommendation applies to, Cloud Advisor uses the billed usage for performance units and storage.

How to implement

Do one of the following:

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow, manual deletion, or manual attachment), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).
Delete unattached boot volumes

The Delete unattached boot volumes recommendation indicates that unattached boot volumes exist in your tenancy. Attaching or deleting unattached boot volumes reduces costs.

Recommendation generation logic

Cloud Advisor scans the tenancy once per day for boot volumes that are not attached to any compute resource. If the volume is unattached for fourteen consecutive days, Cloud Advisor recommends that you delete it.

Savings calculation

Cloud Advisor estimates how much money would be saved by deleting the unattached boot volume. For the resource that the recommendation applies to, Cloud Advisor uses the billed usage for performance units and storage.

How to implement

Do one of the following:

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow, manual deletion, or manual attachment), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).
Downsize underutilized ADW and ATP databases

The Downsize underutilized ADW and ATP databases recommendation indicates that more CPUs are allocated to autonomous databases than you need. Reducing the number of CPUs allocated to your databases saves you money.

Recommendation generation logic

Cloud Advisor identifies low average CPU utilization the last seven days for Autonomous Database for Analytics and Data Warehousing and Autonomous Database for Transaction Processing and Mixed Workloads databases. A low average CPU utilization is one that indicates the utilization to be less than 30% of the allocated CPUs.

Savings calculation

For the resource's compute costs, Cloud Advisor multiplies the billed usage for ADB Compute by the unit price, and then divides that value in half.

How to implement

Do one of the following:

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow or manual CPU adjustment), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).
Downsize underutilized compute instances

The Downsize underutilized compute instances recommendation indicates that some compute instances are bigger than needed. Implementing this recommendation saves you money without degrading performance.

Recommendation generation logic

Cloud Advisor gathers the CPU, Memory, and Network utilization over the selected interval (default seven days) to see if the usage is below the thresholds as specified by the recommendation profile. Based on the methodology of the recommendation profile, Cloud Advisor either uses the Average or P95 statistic for the CPU utilization during evaluation. The network utilization value is computed based on the max of the network bytes in-rate and network bytes out-rate over the network bandwidth in Gbps.

The recommendation is generated only if a smaller instance is available to support the number of VNICs in use.

To customize the logic for this recommendation, see To edit a global recommendation override.

Supported shapes for Downsize underutilized compute instances

Following are the shapes supported for Cloud Advisor scans using Downsize underutilized compute instances:

  • VM.Optimized3.Flex
  • VM.Standard.B1.1
  • VM.Standard.B1.16
  • VM.Standard.B1.2
  • VM.Standard.B1.4
  • VM.Standard.B1.8
  • VM.Standard.E2.1
  • VM.Standard.E2.2
  • VM.Standard.E2.4
  • VM.Standard.E2.8
  • VM.Standard.E3.Flex
  • VM.Standard.E4.Flex
  • VM.Standard1.1
  • VM.Standard1.16
  • VM.Standard1.2
  • VM.Standard1.4
  • VM.Standard1.8
  • VM.Standard2.1
  • VM.Standard2.16
  • VM.Standard2.2
  • VM.Standard2.24
  • VM.Standard2.4
  • VM.Standard2.8

Savings calculation

For the resource that the recommendation applies to, Cloud Advisor subtracts the recommended OCPU from the current OCPU. It then multiplies the result by the billed usage and by the unit price. That value is divided by the current OCPU.

How to implement

Note

  • Select a good time for updating the instances. Because instances require reboots after resizing, we recommend that you resize the instances at a time that does not disrupt your users. It usually takes less than five minutes to reboot an instance, and you can track the progress for each instance by monitoring the associated work request.
  • Downsizing an instance reduces the number of OCPUs, the amount of memory, the network bandwidth, and the maximum number of VNICs for the instance. The instance's public and private IP addresses, volume attachments, and VNIC attachments do not change.

Do one of the following:

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow or manual instance adjustment), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).
Downsize underutilized load balancers

The Downsize underutilized load balancers recommendation indicates that some load balancers are configured with higher minimum bandwidth than needed. Implementing this recommendation saves you money without degrading performance as the minimum bandwidth assigned determines the minimum amount that would be charged for the load balancer.

Recommendation generation logic

Cloud Advisor detects if the average of the maximum values for peak bandwidth usage over the evaluation period is less than the threshold percent of minimum bandwidth as specified by therecommendation profile. For the identified load balancers, Cloud Advisor suggests decreasing the minimum bandwidth to the current peak bandwidth utilization value. The maximum bandwidth value remains the same.

For dynamic shapes, we recommend converting to flexible shape where the current bandwidth of dynamic shape is selected as the maximum bandwidth value. The minimum bandwidth is set to the peak bandwidth utilization value as previously mentioned.

To customize the logic for this recommendation, see To edit a global recommendation override.

Savings calculation

  • If a load balancer with dynamic shape is recommended to be converted to a flexible shape, Cloud Advisor subtracts the recommended minimum bandwidth from the current minimum bandwidth. It then multiplies the result by the billed usage and by the unit price per bandwidth of the current dynamic shape. If the recommended minimum bandwidth is less than or equal to 100 Mbps, that number is multiplied by 0.35 to offset the cost difference between dynamic and flexible shapes.
  • If the recommendation is generated for a load balancer with a flexible shape, Cloud Advisor subtracts the recommended minimum bandwidth from the current minimum bandwidth. It then multiplies the result by the billed usage and unit price of a single flex bandwidth. The billed usage of a single flex bandwidth is calculated by dividing billed usage by the current minimum bandwidth.

How to implement

Note

  • This recommendation changes the load balancer's minimum bandwidth only. The maximum bandwidth remains unchanged.
  • Changing the bandwidth size of the load balancer requires resetting all existing sessions of the load balancer. If you switch to a flexible shape, you cannot revert to a dynamic shape. If the fixed shape is billed at a discounted rate, the estimated savings values might overstate potential savings. Before making changes, work with your sales representative to understand your billing and discount options.

Do one of the following:

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow or manual adjustment, either from dynamic to flexible shape or adjustment of existing flexible shape), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).
Enable monitoring on compute instances

The Enable monitoring on compute instances recommendation indicates that metrics for some running compute instances aren't currently visible to Cloud Advisor. For example, CPU utilization data can't be tracked. Implementing this recommendation makes compute instance metrics visible to Cloud Advisor, allowing recommendations to be provided to you.

Recommendation generation logic

Cloud Advisor identifies compute instances that don't have Monitoring enabled.

How to implement

Manually enable monitoring (this recommendation cannot be implemented using Cloud Advisor). To determine what needs to be changed to enable monitoring, examine the instance. For detailed instructions, see Enabling Monitoring for Compute Instances.

Note

Possible reasons why a compute instance does not have Monitoring enabled:
  • The instance's image does not support monitoring.
  • The instance's image supports monitoring, but the Oracle Cloud Agent is either disabled or not installed. For manual installation instructions, see Installing the Oracle Cloud Agent Software.
  • A service gateway does not exist for that virtual cloud network (VCN).

Status changes

When you implement the recommendation (manually enabling monitoring), its status changes from Pending to Implemented. The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours). If the instance stops running or is deleted, then the recommendation is deleted.

Enable object lifecycle management

The Enable object lifecycle management recommendation indicates that no lifecycle policy rules exist for a Object Storage bucket in your tenancy.

Recommendation generation logic

Cloud Advisor queries the EnabledOLM metric to determine whether Object Lifecycle Management is enabled for buckets in your tenancy. For more information, see Object Storage Metrics. If Object Lifecycle Management is not enabled for the bucket, Cloud Advisor recommends that you enable it.

Savings calculation

  • For each Object Storage bucket, Cloud Advisor multiplies the billed usage by the unit price and then takes 45 percent of that value. Cloud Advisor uses 45 percent because of overall customer trends. These trends indicate that over time, buckets with lifecycle management enabled trend toward a balance of 50 percent archive storage and 50 percent standard storage. Therefore, Cloud Advisor estimates that the current standard storage for a bucket will, over time, convert to a balance of 50 percent archive storage and 50 percent standard storage.
  • The actual ratio varies based on the bucket's purpose. Although Cloud Advisor estimates a 45 percent savings, the savings could be as high as 90 percent or as low as 0.

How to implement

Note

  • To use Object Lifecycle Management, you must first authorize the Object Storage service to archive and delete objects on your behalf. See Service Permissions for more information.
  • After objects are moved to Archive Storage, you must first restore objects to access them.
  • You can also use Object Lifecycle Management to:
    • delete all objects in a bucket or objects that match the names filters that you specify, or
    • delete uncommitted or failed multipart uploads.
  • Cloud Advisor does not provide cost savings estimates for these deletions.
Do one of the following:

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow or manual creation), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes at the next scan of your tenancy (typically within 24 hours), assuming that the metric is emitted.

Performance Recommendations

Performance recommendations find overutilized resources and recommend adjustments.

Enable performance auto-tuning for block volumes

The Enable performance auto-tuning for block volumes recommendation indicates that a block volume is using suboptimal performance settings. Implementing this recommendation improves performance of the volume. The auto-tune feature automatically shifts performance between lower cost, balanced, and higher performance as necessary to optimize utilization of each volume. With the auto-tune feature enabled, you do not need to continually manage volume resources.

For more information about the auto-tune feature, see Auto-tune Volume Performance.

Recommendation generation logic

Cloud Advisor identifies block volumes that don't have the auto-tune performance feature enabled.

How to implement

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow or manually enabling auto-tuning), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).
Enable performance auto-tuning for boot volumes

The Enable performance auto-tuning for boot volumes recommendation indicates that a boot volume is using suboptimal performance settings. Implementing this recommendation improves performance of the volume. The auto-tune feature automatically shifts performance between lower cost, balanced, and higher performance as necessary to optimize utilization of each volume. With the auto-tune feature enabled, you do not need to continually manage volume resources.

For more information about the auto-tune feature, see Auto-tune Volume Performance.

Recommendation generation logic

Cloud Advisor identifies boot volumes that don't have the auto-tune performance feature enabled.

How to implement

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow or manually enabling auto-tuning), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).
Rightsize compute instances

The Rightsize compute instances recommendation indicates that a compute instance doesn't have the shape needed for optimal performance. Implementing this recommendation improves compute instance performance.

Recommendation generation logic

Cloud Advisor gathers the CPU and Memory utilization over the selected interval (default seven days) to see if the usage is above the thresholds as specified by the recommendation profile. Based on the methodology of the recommendation profile, Cloud Advisor either uses the Average or P95 statistic for the CPU utilization during evaluation.

For the identified instances, Cloud Advisor suggests new shapes with higher configurations.

To customize the logic for this recommendation, see To edit a global recommendation override.

How to implement

Note

Before rightsizing compute instances, confirm sufficient service limits. If needed, request a service limit increase.

Do one of the following:

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow or manual adjustment), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).
Rightsize load balancers

The Rightsize load balancers recommendation indicates that a load balancer doesn't have the bandwidth or shape needed for optimal performance. Implementing this recommendation improves load balancer performance.

Recommendation generation logic

Cloud Advisor detects if the average of the maximum values for peak bandwidth usage over the evaluation period is more than the threshold percent of maximum bandwidth as specified by the recommendation profile. For the identified load balancers, Cloud Advisor suggests increasing the maximum bandwidth value by 25%. The minimum bandwidth value remains the same.

For dynamic shapes, we recommend converting to flexible shape where the current bandwidth of dynamic shape is selected as the minimum bandwidth value. The maximum bandwidth is set to be 25% higher than the current bandwidth of the dynamic shape.

To customize the logic for this recommendation, see To edit a global recommendation override.

How to implement

Note

Before rightsizing load balancers, confirm sufficient service limits. If needed, request a service limit increase.

Do one of the following:

Status changes

When you implement the recommendation (Cloud Advisor fix-it flow or manual adjustment, either from dynamic to flexible shape or adjustment of existing flexible shape), its status changes from Pending to Implemented. The time needed to reflect this status change depends on the method used to implement the recommendation.

  • Cloud Advisor fix-it flow: The status changes when the work request completes.
  • Manual implementation: The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).

High Availability Recommendations

High availability recommendations provide hardware failure best practices to ensure the resilience of your solution.

Improve fault tolerance

The Improve fault tolerance recommendation indicates that all virtual machine (VM) instances in the indicated compartment are clustered in a single fault domain. Implementing this recommendation improves availability of your VMs across fault domains.

Recommendation generation logic

Cloud Advisor generates this recommendation when all of a compartment's VMs are in a single fault domain.

How to implement

Manually move instances to spread them across multiple fault domains (this recommendation cannot be implemented using Cloud Advisor). For best practices, see Fault Domains. For instructions, see Editing the Fault Domain for an Instance.

Status changes

When you implement the recommendation (manually moving instances), its status changes from Pending to Implemented. The status changes after the next Cloud Advisor scan for your tenancy (typically within 24 hours).