What’s New

Version 8.13.0

Container Resource Management

• Scale Actions for GenAI Large Language Model (LLM) Inference Workloads

  For GenAI large language model (LLM) inference workloads that use GPU resources and are deployed in a Kubernetes cluster, Turbonomic can now generate workload controller scale actions to maintain Service Level Objectives (SLOs) for these workloads.

  For more information, see Scale Actions for GenAI LLM Inference Workloads.

• Node Provision and Suspension Actions for Google Kubernetes Engine (GKE) Clusters

  For nodes in a Google Kubernetes Engine (GKE) cluster, Turbonomic can now generate node provision actions to address workload congestion or meet application demand, and node suspension actions to improve infrastructure efficiency.

  For more information, see Node Provision and Suspension Actions.

  To execute these actions, Turbonomic must be able to connect to your GKE cluster through a Kubernetes or Red Hat OpenShift target and a Google Cloud target. For the Google Cloud target, the service account requires the following permissions:

  • compute.instanceGroupManagers.update

  • compute.instanceGroups.get

  • container.clusters.get

  • container.clusters.update

  For the full list of permissions, see Permissions for the Google Cloud Target.

• Enhancements for Container Pod Move Actions

  • This release enhances the mechanism for handling pod-to-pod affinity and anti-affinity. With this enhancement, Turbonomic can now generate more accurate container pod move actions that comply with affinity and anti-affinity rules, while continuously driving the target environments toward the desired state. To review these actions in Action Center, look for container pod move actions with an action category of 'Compliance' and a risk of 'Unsatisfied Affinity Constraints' or 'Unsatisfied Anti-Affinity Constraints'.

    For more information, see Pod Affinity and Anti-affinity Rules.

  • When pod move actions execute, the pod clone process now copies and persists labels so that label constraints for the pods comply with any label-specific constraints that exist in the container platform cluster. This enhancement ensures that pod move actions can be successfully completed.

  • For Kubernetes or Red Hat OpenShift environments that have Security Context Constraints (SCC) in use, SCC support is now enabled by default. You no longer need to specify sccsupport="*" in Kubeturbo deployments to enable container pod move actions.

• Improved Management of Prometheus Targets

  To align with industry terminology, the category name for the Prometheus target has changed from 'Custom' to 'Observability'.

  Turbonomic collects application metrics from Prometheus servers through the Prometurbo agent that you deploy to OperatorHub in Red Hat OpenShift. To help you deploy and manage Prometurbo with ease, this release introduces the following improvements.

  • In the user interface, when you navigate to Settings > Target Configuration > New Target > Observability > Prometheus, you will now see a guide that walks you through the Prometurbo deployment process, including the settings that you need to configure directly in OperatorHub. You do not need to configure any settings in the user interface.

    Note:

    The Prometurbo deployment guide is also available in the Turbonomic documentation. For more information, see Prometheus.

    The Prometurbo deployment guide in GitHub is no longer maintained and will be retired. Always refer to the Turbonomic documentation for the latest information.

  • After the deployment, Prometurbo automatically adds itself as a Prometheus target in the user interface, in Settings > Target Configuration. When you view the details for a target, you will now see read-only information about your Prometurbo deployment. To make changes, update the Prometurbo configuration in OperatorHub.

Cloud Resource Management

• Enhancements for AWS VMs Running EC2 GPU Instance Types

  • Support for Additional NVIDIA GPU Metrics

    Turbonomic now discovers the following NVIDIA GPU metrics for AWS VMs running supported EC2 GPU instance types:

    • GPU FP16, FP32, and FP64

    • GPU Tensor

    • GPU Memory BW

    Turbonomic uses these metrics to generate VM scale actions that optimize performance and costs. You can view these metrics in the Capacity and Usage and Multiple Resources charts.

    For more information, see Support for AWS EC2 GPU Instance Types.

    To enable the discovery of these metrics, configure NVIDIA Data Center GPU Manager (DCGM). For more information, see Linux Configuration for NVIDIA GPU Metrics (DCGM).

  • Ignore NVIDIA GPU Compute Capability

    This constraint is a setting that you can choose to turn on in automation policies for AWS VMs. When this setting is turned on, scale actions that change the GPU compute capability of VMs are allowed to execute in Turbonomic.

• Newly Supported AWS Instance Families and Regions

  This release introduces support for the following AWS instance families and regions.

  Instance families:

  • C6id, C6in, C7a, C7g, C7gd, C7gn, C7i

  • Hpc7g

  • I4g, I4i

  • M6id, M6idn, M6in, M7a, M7gd, M7i, M7i-Flex

  • R6a, R6id, R6idn, R6in, R7a, R7gd, R7i, R7iz

  • U-1 (High Memory) – u-18tb1 and u-24tb1 only

  Regions:

  • ap-south-2 – Asia Pacific (Hyderabad)

  • ap-southeast-4 – Asia Pacific (Melbourne)

  • ca-west-1 – Canada West (Calgary)

  • eu-central-2 – Europe (Zurich)

  • eu-south-2 – Europe (Spain)

  • il-central-1 – Israel (Tel Aviv)

  • me-central-1 – Middle East (UAE)

• Support for AWS Standard Data Exports (CUR 2.0)

  The AWS Billing target can now retrieve billing data from a standard data export (CUR 2.0) that you set up in the AWS Billing and Cost Management console.

  If you previously set up a legacy CUR export for use with the AWS Billing target and now want to switch to the standard data export:

  1. Set up a standard data export. For more information, see Setting Up a Standard Data Export (CUR 2.0).

  2. In the Turbonomic user interface, remove your existing AWS Billing target and then add a new one. In the new target, specify the S3 bucket name, S3 path prefix, and S3 bucket region of the standard data export.

  Note:

  You can continue to use a legacy CUR export until further notice.

• AWS RDS Data in Discount Charts

  For AWS RDS Reserved Instances (RIs), utilization and coverage data is now available in the Discount Utilization and Discount Coverage charts.

  Note:

  Currently, utilization data is not considered in database server scaling recommendations. In addition, Turbonomic does not generate actions to purchase RDS RIs.

• Support for Azure Zone-redundant Storage (ZRS)

  Turbonomic now discovers Azure zone-redundant storage (ZRS) with Premium SSD and Standard SSD managed disks.

  Turbonomic can generate volume scale actions for ZRS to optimize performance and costs. If a ZRS is currently unattached, Turbonomic immediately generates a volume delete action as a cost-saving measure. Savings and investments associated with these actions are tracked in the Savings, Cumulative Savings, Investments, and Cumulative Investments charts.

• Azure Billing Target Support for Azure Government

  The Azure Billing target can now discover and monitor billing data for Azure Government workloads. To ensure seamless monitoring of billing data, follow the guidelines outlined in Support for Azure Government.

  With this new capability, the Azure Billing target is now the only target that Turbonomic requires to monitor billing data for both global Azure and Azure Government workloads. The Microsoft Enterprise Agreement target is no longer required and has therefore reached end of support. If you have an existing Microsoft Enterprise Agreement target in the Settings > Target Configuration page in the user interface, remove the target immediately and then set up an Azure Billing target. For more information about end of support and next steps, see Notice: Microsoft Enterprise Agreement Target.

• Azure Volume Delete Action Enhancement

  Turbonomic now checks the DiskState property of Azure volumes to accurately determine their attachment state. If the DiskState is Unattached, Turbonomic generates an action to delete the volume.

• Scale Actions for Google Cloud VMs Running Custom Machine Types

  Turbonomic can now generate scale actions for Google Cloud VMs running custom machine types to optimize performance and costs. Currently, Turbonomic supports the N1, N2, and N2D custom machine series.

• Historically-informed Action Generation for Google Cloud VMs

  Turbonomic can now generate VM scale actions informed by historical utilization data on the same day a new Google Cloud target is added.

  To take advantage of this feature, add at least one Google Cloud target. The service account for the target requires the logging.logEntries.list permission. For the full list of permissions, see Permissions for the Google Cloud Target.

  Note:

  Currently, VM scale actions consider historical utilization of all monitored resources, except I/O throughput.

• Tracking of Externally-executed Cloud Actions

  Turbonomic now tracks actions for cloud workloads that you executed externally. For example, you may have executed these actions through your cloud provider.

  Use the following charts to view relevant data:

  • Savings, Cumulative Savings, Investments, and Cumulative Investments

    The total amounts shown in these charts now include savings or investments associated with externally-executed actions.

  • Executed Actions - Succeeded

    For externally-executed actions, the Execution column shows 'Executed Externally'.

On-prem Resource Management

• Sustainability Metrics in VM Move and Host Suspend Actions

  For on-prem environments, Turbonomic now shows the projected change in carbon footprint and energy usage in the action details for VM move and host suspend actions. For host suspend actions, a new Energy Usage column in the Action Center summary also shows energy usage data for hosts. For more information, see Sustainability Metrics in VM Move and Host Suspend Actions.

• Stitching Between Datadog and vCenter Targets

  When both Datadog and vCenter targets are added, VM entities will be stitched and a complete supply chain will be available.

• Discovery of VM Virtual Disks for Nutanix Targets

  For Nutanix targets, VM virtual disks are now discovered and displayed as volume entities in the supply chain. The new Nutanix probe property (enableVolumeDiscovery) is enabled by default.

• Handling of Duplicate Storage Targets

  This release implements a duplicate target check for storage targets. If you added duplicate targets, you will now see discovery errors to notify you to remove the duplicates.

Orchestration and IT Management

• Support for ServiceNow Washington DC Release

  Turbonomic Actions for ServiceNow is an integration that brings the power of Turbonomic to your ServiceNow® Change Management application. Turbonomic Actions 1.1.11 now supports the ServiceNow Washington DC release. You can download this integration from the ServiceNow store. For more information, see the Turbonomic Actions for ServiceNow documentation.

• Enhancements for Flexera Targets

  You can now target FlexNet Manager Suite 2023 R2 (on-premises) with Turbonomic to discover software license assets in the same manner as the existing Flexera One (SaaS) integration. By discovering software license assets, Turbonomic can more intelligently place or resize workloads to ensure compliance with licensing policies. Operations teams no longer need to manually define these constraints. Application owners will see optimized license usage in Flexera One and FlexNet Manager by using Turbonomic to optimize license consumption.

  The category name for the Flexera target has changed from "Orchestrator" to "IT Management."

  For more information, see Flexera.

User Interface Management

• Migration of User Interface Pages to the New Design Framework

  The following pages in the user interface now use the new design framework.

  • Settings > Target Configuration

  • Settings > Billing and Costs

  This is part of an ongoing effort to transition the product to the framework, which delivers a high-performance user interface with a more modern look-and-feel.

  To switch to the new framework, click in the navigation bar of the user interface and then turn on the toggle. For more information, see Design Framework for the User Interface.

• Telemetry and Usage Data Sharing

  After upgrading to this version, the Turbonomic telemetry and usage data sharing feature is automatically enabled. Telemetry data, including IP addresses and approximate geolocation, is sent automatically to IBM and/or its sub-processors. The data is only used by IBM for product development and feature improvement. You can opt out of the automatic telemetry data upload at any time by disabling it in the Settings > Maintenance Options > Telemetry Data Sharing page.

  For more information, see Maintenance Options.

API Management

• Programmatic Authentication with Turbonomic APIs by Using OAuth 2.0 Tokens

  This release includes a new OAuth 2.0 feature, which provides an easier way to programmatically authenticate with Turbonomic APIs. You can now use the Turbonomic API to create, list, and delete OAuth 2.0 clients. After you create a client, you can programmatically exchange the client credentials for access tokens.

  For more information, see Authenticating OAuth 2.0 clients with the API.

• Improvement for Searches that Use Regular Expressions

  If you used a regular expression (regex) to search for a group in the user interface and specified * as the first character, the /search API call now performs a regex validation to prevent errors with error code 500.