Reviewing Real-Time Utilization Data

This section describes the available utilization metrics that are visible in the Virtualization Manager and actions you can take to access utilization data from the Virtualization Manager.

Utilization Metrics Visible in Virtualization Manager

The Virtualization Manager displays real-time utilization metrics from the managed systems. Utilization metrics graphically display as meters in a Visualization View compartment or as a row in the table in the Workload View. The meters show CPU utilization, memory utilization, network I/O, and disk I/O (see “Available Utilization Metrics”). Power meters are provided for any physical server, including standalone servers, nPartitions, or server blades. Power meters typically display a warning triangle until a power calibration has been performed for a system. Once calibrated, the power meter value can be displayed. You can link to the power calibration task is when it is present in the meter callout or from the ConfigureCalibrate Power (All Selected Systems)... menu option. For information on the meter callout, see “Meter Callout Information”. For information on Configure menu options, see Configure Menu.




	NOTE: Because power settings are associated with hardware, power meters do not display for VM guests (including HP virtual machines, Microsoft virtual machines, or VMware), complexes, enclosures, Virtual Connect Domains, and Virtual Connect Domain groups.

CPU, memory, network I/O, and disk I/O meter information is obtained via WBEM from the Utilization Provider running on the managed systems, Virtual Machine Management, or other sources depending on the physical or virtual system you are viewing. Power meter data is based on CPU utilization; therefore, CPU utilization data needs to be present (in addition to the power calibration) before power meters can be shown.

The Utilization Provider maintains a running average of the utilization metrics, updating the information every 5 minutes. The average is calculated by comparing the current sample with the sample of 5 minutes earlier, so two samples are required before utilization data is available. This means that it can take up to 10 minutes from the time that the Utilization Provider begins monitoring a system or workload before utilization data is available. Other sources of utilization data have a similar interval for updating the data.

During that interval, the utilization meters are grayed out in the Workload View or have an hourglass icon on them in the Visualization View. If no data is available, dashes (- -) appear over the meter (for example, a system might be disabled). HP SIM status icons can also appear in a meter indicating a status condition, such as a problem collecting utilization data. A warning triangle indicates a problem with collection of the meter data and that details can be found in the meter callout (see “Meter Callout Information” for information on meter callouts).

The Virtualization Manager display is updated at regular intervals. This data is a maximum of 5 minutes old.

Accessing Utilization Data

You can access utilization data from the Visualization View and Workload View.

From the Visualization Tab

Click the Visualization tab to display the Visualization View (the default view when you first access the Virtualization Manager). In a system compartment, hover over the meter callout icon . This action displays callout information for the all available utilization metrics, including any error conditions. From the callout information box (see “Meter Callout Information”), you can view more information, including real-time data and historical data. Click the link next to the metric to display the Capacity Advisor Profile Viewer. For information on the Capacity Advisor Profile Viewer, see the HP Capacity Advisor Version 4.1 User's Guide.

From the Workload tab

Click the Workload tab to display the Workload View. The utilization metrics are displayed in table rows on the main page. You can view data for any metric by clicking its meter or corresponding percentage estimate, both of which displays the Capacity Advisor Profile Viewer.

If no data has been collected yet, you can use the Capacity Advisor to begin collecting data.

Available Utilization Metrics

The following types of utilization metrics are shown on the Visualization View and Workload View:

CPU Utilization Percentage of total CPU resources in use. For a complex, virtual partition server, enclosure, VC domain, VC domain group, or Serviceguard cluster compartment, the aggregate CPU utilization is calculated as the average utilization of its subcompartments. CPU meters shown in compartments that do not have CPUs are an aggregation.
Memory Utilization Percentage of total memory resources in use, in GB (defined as 2³⁰ bits). The memory utilization shown for an entire system is not the same as the sum of the memory utilization of all of the workloads it contains. In the data for the entire system, this metric includes dynamic buffer cache and system memory that is not directly attributable to any specific process and therefore is not included in any workload.
For a complex, virtual partition server, enclosure, VC domain, VC domain group, or Serviceguard cluster compartment, the aggregate memory utilization is calculated as the average utilization of its subcompartments. Memory meters shown in compartments that do not have memory are an aggregation.
Network I/O Network throughput (transmitted plus received), in one of Kb/s (10³ bits per second) Mb/s (10⁶ bits per second) Gb/s (10⁹ bits per second). Current utilization determines the units to be used. On managed systems running HP-UX or Microsoft Windows, the maximum length of the Network I/O meter is calibrated to a theoretical maximum based on the network interface type. On Windows managed systems where data is collected by way of the Agentless Data Collector or on Linux managed systems, the maximum is calibrated to a “high-water mark” (the highest value seen on this system so far) that is tracked by the Utilization Provider. For VMM, the maximum is first retrieved from the setting obtained from ConfigureEdit Network and Disk I/O Capacity…. If the maximum is not set there, the last 5 hours worth of VMM data is used to calculate a high-water mark.
For a complex, virtual partition server, enclosure, VC domain, VC domain group, or Serviceguard cluster compartment, the aggregate network I/O utilization is calculated as sum of the utilization of its subcompartments. Note that Network I/O meters shown in compartments that do not have network I/O are an aggregation.
Disk I/O Disk I/O throughput (stored plus retrieved), in one of KB/s (10³ bits per second) MB/s (10⁶ bits per second) GB/s (10⁹ bits per second). Current utilization determines the units to be used. The maximum length of the disk I/O meter is calibrated to a high-water mark (the highest value seen on this system so far) that is tracked by the Utilization Provider. For VMM, the maximum is first retrieved from the setting obtained from ConfigureEdit Network and Disk I/O Capacity…. If the maximum is not set there, the last 5 hours worth of VMM data is used to calculate a highwater mark.
For a complex, virtual partition server, enclosure, VC domain, VC domain group, or Serviceguard cluster compartment, the aggregate disk I/O utilization is calculated as sum of the utilization of its subcompartments. Disk I/O meters shown in compartments that do not have disk I/O are an aggregation.
Power The power meter shows the power utilization of a compartment in Watts (where supported), based on calibration information designated from ConfigureCalibrate Power (All Selected Systems)... or alternatively, from the Calibrate Power link in the meter callout. When CPU utilization data is available, power meters are provided for any physical server, including standalone servers, nPartitions, or server blades. Power meters typically display a warning triangle until a power calibration has been performed for a system. Once calibrated using either of the above methods, the power meter value can be displayed.

Meter Callout Information

Meter callout information includes:

Current utilization data for CPU, memory, network, and disk; plus the ability to view historical data, for example, by linking to Capacity Advisor to display a profile page.
Information about power consumption for any physical server, including standalone servers, nPartitions, or server blades.
Collection interval; which is typically in 5 minute intervals for WBEM/VMM data, but can vary for agentless data collection.
Source of the meter data, indicating whether the data comes from the utilization WBEM provider, VMM, or via agentless data collection.
Last-updated time of the data collection.
Error messages (for example, if there was a problem collecting utilization data from the server), a suggested resolution, and an optional Retry link.

The following figure shows meter callout information for a server blade running on HP-UX.

If the compartment shows a managed workload, links appear that let you view real time and historical information from the Global Workload Manager (gWLM). For example: