IBM PowerVM

Prerequisites

• Supported Hardware Management Console (HMC) versions:

  • HMC 10: Power8, Power9, or Power10

  • HMC 9: Power8 or Power9 only; Power7 LPARs and Systems are ignored.

• Compute Long Term Monitor Metrics (Compute LTM) must be enabled on each system in the HMC. You can enable this function either manually by using the HMC API or in the Turbonomic user interface (UI) when you add or edit the target.

  • To manually enable Compute LTM by using the HMC for each Power System that you want Turbonomic to discover, configure the following setting by using the HMC API. You cannot enable the setting through the HMC UI.

    • Set the ComputeLTMEnabled property to true for all partitions.

      Take care not to confuse this property with the LongTermMonitorEnabled property. ComputeLTMEnabled and LongTermMonitorEnabled are separate properties and ComputeLTMEnabled is the property that you must set to true to enable metrics gathering.

    • For more information, see the Management Console Performance and Capacity Monitoring Preferences for your specific system.

  • Alternatively, when you add or edit the target in the Turbonomic UI, select Automatically enable HMC Long Term Metric collection on Power Systems to enable Compute LTM on all capable systems during discovery. This option is not enabled by default and it requires higher HMC account permissions, which are detailed in the user account prerequisites.

• Optionally, you can enable Energy Monitoring to view the power and energy metrics on Turbonomic. You can enable this function either manually by using the HMC API or in the Turbonomic UI when you add or edit the target.

  • To manually enable Energy Monitoring for each system in the HMC, enable performance collection from HMC UI:

    • The value for the EnergyMonitoringCapable attribute must be true. This read-only attribute determines whether the managed system supports energy data collection.

    • Enable energy monitoring by setting the EnergyMonitorEnabled property to true.

    • For more information, see the Management Console Performance and Capacity Monitoring Preferences for your specific system.

  • Alternatively, when you add or edit the target in the Turbonomic UI, select Automatically enable energy monitoring on energy monitoring capable Power Systems to enable energy monitoring on all capable systems during discovery. This option is not enabled by default and it requires higher HMC account permissions, which are detailed in the user account prerequisites.

• An HMC user account that Turbonomic can use to connect to the PowerVM target.

  The HMC user account must have the following roles:

  • The hmcviewer role is required to discover and view metrics. This role cannot edit HMC settings.

  • To use the Automatically enable HMC Long Term Metric collection on Power Systems or Automatically enable energy monitoring on energy monitoring capable Power Systems options in the Turbonomic UI, the user account must have one of the following two permissions:
    • The hmcoperator role
    • The Manage Partition Utilization Data task role

  The user account must also be enabled for remote access by selecting the Allow remote access via the web option in the HMC user properties.

  To enable this option, select HMC Management in the HMC navigation panel and click Manage User Profiles and Access. Select the user profile and click the User, Modify, menu option. In the Modify User dialog, click the User Properties... button and select the Allow remote access via the web checkbox. Click OK, OK to save the change. For more information, see HMC Version 7 and later: How to Access the HMC GUI Remotely.

• The HTTPS TCP port 443 from Turbonomic to the HMC instance must allow connections from Turbonomic to the HMC.

IBM PowerVM target configuration details

Before you configure an IBM PowerVM target, review the following considerations:

• Turbonomic recommends Resize LPAR actions only.

• Stitching to entities above the LPAR/VM or below the System/Host is not currently supported.

• Where multiple HMCs are managing the same Power Systems (Clustered/HA configuration) add only one of the HMC nodes as a target to Turbonomic.

• Plans are not currently supported. Planning features are being considered in a future release.

• Processor pools are not currently supported.

• Reporting that uses ThoughtSpot is not currently supported.

• The Details tab requires manual configuration for static group scopes.

• VIOS LPARs are not discovered currently.

• In a few cases, actions to resize LPAR PU and VP resources might result in recommendations for PU allocations above VP capacity and vice versa.

Adding IBM PowerVM targets

spec:
  powervm:
    enabled: true

1. Click Settings > Target Configuration.

2. On the Target configuration page, click Add Target.

3. On the Select target page, click IBM PowerVM.

4. In the side panel, review the connection requirements and then click Connect Target.

5. Configure the following settings:

   • Display name

     Specify a name that uniquely identifies this connection.

     The name that you provide here is also used for the Data Center and Cluster name to which Systems managed by this HMC belong.

   • Address

     Specify the IP/Domain of the PowerVM HMC.

   • Port number

     Specify the port number that Turbonomic uses to connect to remote management. By default, the port is 443 for HTTPS.

   • Username

     Specify the username of the account Turbonomic uses to connect to the target.

   • Password

     Specify the password of the account Turbonomic uses to connect to the target.

   • Enable HMC Long Term Metric collection on Power Systems

     If you select this option, Turbonomic automatically enables Compute-Long Term Monitor metrics (LTM) on each power system that the HMC manages. By default, Turbonomic does not make any changes on HMC; however, Compute-LTM must be enabled on each Power System either automatically (enable this option) or manually.

     Compute-LTM is a prerequisite for metrics collection.

     To enable this option, the user account that is used by Turbonomic must have Manage Partition Utilization Data as a Task Role in HMC.

   • Enable energy monitoring on energy monitoring capable Power Systems

     If you select this option, Turbonomic automatically enables energy monitoring for hosts that are energy monitoring capable and enabled on each power system that the HMC manages.

Entity mapping

After validating your targets, Turbonomic updates the supply chain with the entities that it discovered. The following table describes the entity mapping between the target and Turbonomic.

Monitored resources

Turbonomic monitors the following resources:

• Virtual Machine

  • Processing unit (PU)

    For PowerVM, processing unit is a unit of measurement for shared processing power across one or more virtual processors. One shared processing unit on one virtual processor accomplishes approximately the same work as one dedicated processor.

    PUs can be a decimal value in a VM setting. A VM that has a capacity of 0.5 PU means that 0.5 PU is always readily available for that VM to use, and that amount can only be used for that VM.

  • Virtual processor (VP)

    For PowerVM, virtual processor (VP) is the guest OS representation of the capacity of the VM. VPs have a one-to-one correlation with PUs. Therefore ,1 VP that is utilized in a VM equates to 1 PU being used. However, VPs do not have to be equal to PUs on a system. VPs are a virtual representation. Therefore, they can be overprovisioned.

    For example, a VM can have 0.5 PUs but 3 VPs. In this scenario, a VM’s capacity is 0.5 PU but it can use more than its assigned capacity; the max it can use is the number of VPs. The number of VPs must never exceed 20x the value of PU, therefore the highest utilization a VM can be at is 2000% of its PU capacity.

    Note:

    Only shared-uncapped VM utilization can go up to 2000% of its PU capacity; the utilization for all other VM types cannot go above 100% of its PU capacity.

• Host

  • Energy

    Energy is the measurement of electricity used by a given entity over a period of time, expressed in watt-hours (Wh).

  • Carbon footprint

    Carbon footprint is the measurement of carbon dioxide equivalent (CO2e) emissions for a given entity. Turbonomic measures carbon footprint in grams.

  • Power

    Power is the measurement of electricity consumed by a given entity, expressed in watts.

  • Memory (Mem)

    Memory is the measurement of memory that is reserved or in use.

  • Virtual processor (VP)

    For PowerVM, virtual processor (VP) is the guest OS representation of the capacity of the VM. VPs have a one-to-one correlation with PUs. Therefore ,1 VP that is utilized in a VM equates to 1 PU being used. However, VPs do not have to be equal to PUs on a system. VPs are a virtual representation. Therefore, they can be overprovisioned.

    For example, a VM can have 0.5 PUs but 3 VPs. In this scenario, a VM’s capacity is 0.5 PU but it can use more than its assigned capacity; the max it can use is the number of VPs. The number of VPs must never exceed 20x the value of PU, therefore the highest utilization a VM can be at is 2000% of its PU capacity.

    Note:

    Only shared-uncapped VM utilization can go up to 2000% of its PU capacity; the utilization for all other VM types cannot go above 100% of its PU capacity.

• Datacenter

  The Datacenter entity is not monitored but is used in the supply chain to represent each PowerVM HMC and the Hosts and VMs associated with it.

  • Power

    Power is the measurement of electricity consumed by a given entity, expressed in watts.

Actions

Turbonomic supports the following actions:

• Virtual Machine

  Resize PU or VP (IBM PowerVM)

  Resize PU (processing units) or VP (virtual processor) to optimize LPAR processing unit allocation and virtual processor capacity based on historical demand collected from the HMC.

  When Turbonomic generates an action, the action details include a list of policies that constrain the action recommendation. In this list, policy names that begin with IBM PowerVM: indicate policies that implement the constraints suitable for PowerVM. You can click to navigate to the policy definition. You can see the policy definition, but you cannot edit it.

  The policies that Turbonomic generates affect Turbonomic actions in the following ways:

  Entity	Policy Impact on Actions
  Virtual Machine	Turbonomic does not support native execution of Resize PU (processing units) and Resize VP (virtual processor) actions. However, actions can be executed using external automation platforms through automation workflows such as Action Scripts or Webhooks. The following policy settings affect the resize actions that Turbonomic generates: Min/Max observation period Observation period settings only apply to Resize VP actions. For details, see Aggressiveness and Observation Periods. Increment constant for processing unit Change the increment constant value to use for VM PU resize actions. PUs can be a decimal value in a VM setting. For details, see Increment Constant for Processing Unit. Increment constant for virtual processors Change the increment constant value to use for VM VP resize actions. VPs must be an integer value in a VM setting. For details, see Increment Constant for Virtual Processors.
  Host	Memory The memory that is given to a VM from the host system; what is provided by the host is all that the VM can use. Memory Overprovisioned Percentage Change the percentage of the existing capacity that Turbonomic considers to be 100% of capacity. For details, see Utilization Constraints.