Rightsizing resources

  • Release version: Xanadu
  • Updated August 1, 2024
  • 2 minutes to read
    • Summarize
    Summarized using AI
    This content was generated using new OpenAI-powered functionality. Results are provided on an as is basis and are not guaranteed to be accurate or complete.

    Summary of Rightsizing resources

    The Rightsizing feature in ServiceNow Cloud Cost Management helps customers optimize their cloud resource usage by analyzing consumption data and recommending better-sized resources. These recommendations identify over-provisioned or underused resources that lead to unnecessary costs. Each recommendation includes a confidence rating and predicted savings to assist in decision-making.

    Show full answer Show less

    Customers can schedule Rightsizing jobs to resize selected resources automatically, with integration into ServiceNow Change Management to manage approvals and track changes.

    How Rightsizing Works

    • Rightsizing recommendations are updated whenever billing and usage data refresh.
    • Customers review recommendations on the Rightsizing recommendations page and select resources to include in a Rightsizing job.
    • When creating a job, define the execution schedule and specify the approval type: auto-approval (Standard Change, auto-approved) or manual approval (Normal Change requiring user approval).
    • Once saved, change requests are generated immediately, and the job runs at the scheduled time to resize resources.
    • The system handles resource resizing by stopping, resizing, and restarting resources if needed, with rollback support on failed attempts (AWS only).
    • Rightsizing reports update dynamically to reflect the status of changes, enabling rescheduling of pending, rejected, or failed operations.

    Important Note: Rightsizing on stopped AWS RDS databases is not supported; these databases must be started before resizing.

    Supported Resources and Providers

    • Virtual Machines for AWS, Azure, and Google Cloud Platform (GCP)
    • SQL Databases for Azure and GCP
    • RDS Databases for AWS
    • Storage volumes for AWS Elastic Block Store and Azure Disk

    Recommendations are based on CPU, memory, and network usage metrics to ensure resource adjustments align with actual consumption.

    Confidence Levels in Recommendations

    Each Rightsizing recommendation includes a confidence level to guide customer decisions:

    • High confidence: At least 10 days of usage data and the current and recommended resource families/generations are the same.
    • Medium confidence: Less than 10 days of usage data but the same family/generation.
    • Low confidence: Different current and recommended resource families/generations.

    Customers should consider these confidence levels when evaluating recommendations to ensure appropriate resizing decisions.

    The Rightsizing feature analyzes resource usage to recommend better sizes for resources that are wasting money by being over-provisioned or underused. A confidence rating and predicted savings support each recommendation. Schedule Rightsizing jobs to resize the resources you specify.

    How Rightsizing works

    Flow of the Rightsizing process

    The system updates Rightsizing recommendations each time that billing and usage data are updated.

    Follow this process to define a Rightsizing job:
    1. On the Rightsizing recommendations page, select the resources to rightsize based on your analysis of the recommendations. For more information, see Resize resources with Rightsizing.
    2. Add the resources to a Rightsizing job. The job can be a new or an already-defined one.
    3. Specify the date and time for the job to run.
    4. Specify the type of approval required for the Rightsizing action.

      Rightsizing operations are directly integrated with the ServiceNow Change Management feature.

      • Auto-approval: Generates a Standard Change request and the change request is auto-approved.
      • Manual approval: Generates a Normal Change request and the appropriate user approves the change request.
    5. Save the job.

    When you save the job, the system immediately generates the change requests. Later, at the scheduled time, the system runs the job. The job performs the following operations:

    • For each approved change, resize the resource. For a resource in the ON state, stop the resource, resize it, and then restart it. If the attempt to resize fails, perform a rollback. For more information, see AWS only – Rollback on failed Rightsizing attempts.
    • Update the Rightsizing reports with new recommendations and with approved, successful, pending, rejected, and failed changes.

    For pending, rejected, and failed change requests, you can reschedule the resources into another job.

    Note:
    Rightsizing operation on a stopped AWS Relational Database Service (RDS) database isn’t supported from the AWS provider. If you try to perform Rightsizing on stopped databases, the resize operation fails with the error InvalidDBInstanceState - You can't modify a stopped DB instance. Start the DB instance, and then modify it..

    How Cloud Cost Management generates Rightsizing recommendations

    Cloud Cost Management uses a process that is optimized for each provider.

    Recommendations

    The Cloud Cost Management application can generate recommendations for Virtual Machines (AWS, Azure, GCP), SQL Databases (Azure, GCP), and RDS Databases (AWS). CPU, memory, and network usage metrics are used to generate database rightsizing recommendations for the database resources.

    The Cloud Cost Management application generates recommendations for storage volumes for AWS and Azure providers:
    • Cloud category - AWS Elastic Block Store for Storage Volumes: Service category is storage.
    • Cloud category - Azure Disk for Storage Volumes: Service category is storage.

    Confidence levels in recommendations

    Each recommendation that the system makes to rightsize a resource has an associated confidence level. You consider the confidence level while deciding whether to rightsize a resource. Confidence levels reflect the following factors:
    • High confidence requires the following conditions:
      • The system has 10 or more days of usage data for the resource.
      • The current and recommended family/generation are identical.
    • Medium confidence requires the following conditions:
      • The system has less than 10 days of usage data for the resource.
      • The current and recommended family/generation are identical.
    • Low confidence: The current and recommended family/generation are different.