VMware

VMware VirtualCenter 1.1 Support Documentation


Features | Documentation | Knowledge Base | Discussion Forums

previous Prev   Contents   Last   Next next

Viewing the Performance Tab

The Performance tab is available when you select a farm, host, virtual machine group, or virtual machine object from the inventory panel of the navigation bar.

The Performance tab displays the selected object's information in graph form. Performance views show graphs for four primary attributes: CPU Usage, Memory Usage, Net I/O Usage, and Disk I/O Usage.

Each attribute has a summary graph and one graph per instance of that attribute. For example, the CPU Usage attribute has one graph per CPU. Net Usage has one graph per NIC. Only one set of attribute graphs is viewable at a time.

Performance charts for hosts and virtual machines display in two forms:

  • Summary charts — For each host or virtual machine in a farm or group.
  • Individual charts — For a selected individual resource sharing the panel with thumbnail resource charts for all charted resources: CPU, Memory, NIC, and Disk.

Performance charts for farms and virtual machine groups display the summary charts for each host or virtual machine in their group, respectively.

The Performance tab topics are covered in the following sections:

Viewing Summary Performance Charts

Viewing Summary Performance Charts

The Performance tab of a farm for a virtual machine group object displays a summary chart for each host in the farm or each virtual machine in a virtual machine group.

The summary chart list of the total CPU, memory, disk, and network utilization over a selected time range, along with the average utilization over that same time range. Clicking on the charts links to the Performance tab of the selected host or virtual machine and displays the selected resource as the main chart.

If there are more than four hosts or virtual machines in the farm or virtual machine group, then the First, Prev, Next, and Last buttons are available to shift the view to additional hosts or virtual machines.

Inventory > Farm > Performance Tab

The Performance tab content is described in the following table.

 List Attribute
 Description
 Name
 Host or virtual machine name.
 CPU
 Reports the percentage of CPU being used.
 Memory
 Reports the percentage of usage in MBytes.
 Disks
 Reports the number of bits per second of read and write activity to and from the disk in KBytes per second.
 Network
 Reports the number of bits per second of input and output to each and all network interface (NIC) cards.
 Range
 Selectable range of time to display in the charts. Options include: day, week, month, year.

Viewing Individual Charts

Viewing Individual Charts

The Performance tab for hosts and virtual machines displays four thumbnail charts and one expanded chart of the monitored resource for each selected host or virtual machine.

The thumbnail charts at the top of the panel are the same as the summary charts. Clicking a the summary chart displays an expanded version of that chart in the body of the panel.

Inventory > Host > Performance > CPU Chart

Note: When VirtualCenter server starts, it can take awhile to load the historical statics from the database into memory. While the data is loading: the performance charts are not viewable; the CPU and memory utilization columns in the Hosts and Virtual Machines tabs are empty; and alarms do not fire. All other VirtualCenter functions operate normally.

Performance Panel Legend — View or hide the legend for the charts or not by toggling the Hide legend button. The legend at the bottom of the panel displays a key for:

  • Chart Entries — Maps the line styles to the item they are representing
  • Event Indicators — Triangles on the expanded chart indicate when significant events occurred for the host or virtual machine. Review the Events panel to locate the nature and condition of the event.

Performance Panel Range — Adjust the time increment by selecting from the Range pull-down menu.

Customize Chart Dialog — The Customize Performance Chart provides options that affect the display of the chart. The chart options differ between hosts and virtual machines.

Host customized chart options:

  • All resources:
    • Total utilization [line chart]
    • Total utilization by virtual machine [stacked chart]

      This also provides options Show/Granted and a selectable list of virtual machines to include in the chart data.

  • Disk and network only:
    • Total disk/network I/O
    • Total disk/network I/O by device
    • Total disk/network I/O by virtual machine [stacked chart]

      This option and Individual disk I/O provide an option for Total/Reads/Writes and a selectable list of virtual machines to include in the chart data.

    • Individual disk/network I/O

      This provides and option to specify a device and list the performance data by virtual machine.

Virtual machine chart options:

  • Total utilization [line chart] — All devices
  • Total disk/network by device [stacked chart] — Disk and Network only

    This option and Individual disk I/O provide an option for Total/Reads/Writes and a selectable list of virtual machines to include in the chart data.

  • Individual disk I/O — Disk only

    This provides and option to specify a device in the virtual machine.

Performance Panel Resource Charts

Performance Panel Resource Charts

The performance panel displays the usage of the following resources:

  • CPU — Chart reports the percentage of CPU being used. Calculated as a percentage of shares of the entire CPU availability assigned to the virtual machine.

    The CPU chart displays the percentage of combined CPU usage. If there is more than one CPU the chart distributes the data based on the combined usage at 100%. The legend lists the values as a portion of the 100% total.

    Note: If multiple virtual machines are sharing the same CPU or memory, the usage line might step down or up as additional virtual machines are powered on or off. If there is no contention for the CPU or memory, usage is reported as a flat line.

    The virtual machine CPU and memory usage chart has several lines on it. One each designates the maximum and minimum percentage of the CPU or memory allotted. A third line shows the percent usage of the CPU or memory.

    The CPU chart displays the following information

    • Usage: Percentage of CPU time used by the virtual machine (or by all virtual machines on the host)
    • Guarantee: Percentage of CPU time guaranteed to be available to the virtual machine (not available on the host chart); value is determined by the min/max/shares settings of this VM and all VMs currently running on the host
  • Memory — Chart reports the memory usage in MBytes. Calculated as a percentage of shares of the entire physical plus swap memory on the host assigned to the virtual machine. Virtual machine memory maximum and minimum usage is configurable.

    The display is one chart per host. This does not include NUMA systems.

    The memory chart displays the following information:

    • Active: Memory recently used by the virtual machine (or by all virtual machines on the host)
    • Granted: Memory allocated to the virtual machine (or to all virtual machines on the host)
  • Disk — Chart reports the amount of read and write activity to and from the disk in bytes per second.

    The data is defined by shares of the disk resource per volume. There is no line to show the percentage of the disk space read/write usage assigned to the virtual machine.

    Data is collected on each disk connected to the ESX Server and GSX Server host. This tracks the VMFS traffic and access to the network share. For any disk, the reported data is relative to the host selected in the inventory panel. Traffic with a disk that is due to additional hosts attached to the disk is not included in the chart.

    Host data appears in two aggregate charts, one for read activity and one for write activity. With VMFS systems, separate charts display volume inbound read traffic and volume outbound write traffic. Note logical volumes span partitions.

    Note: Disk charts do not report percentage of occupancy, they do report read/ write activity.

    The disk chart displays the following information:

    • Total: Total I/O
    • Reads: Disk reads, network traffic received
    • Writes: Disk writes, network traffic transmitted
  • Network — Chart reports the number of bits per second of input and output to each and all network interface (NIC) cards.

    Host aggregate chart reports all the network input and output per host. Network input and network output usage are given separate charts reporting the usage of the physical NIC based on the configuration of the virtual machines on that host.

    Virtual machine aggregate chart shows all the input and output activity per virtual machine. The detail charts show the activity for each NIC for the virtual machine.

    The network chart displays the following information:

    • Total: Total I/O
    • Reads: Disk reads, network traffic received
    • Writes: Disk writes, network traffic transmitted
Configuring Performance Measurement Intervals

Configuring Performance Measurement Intervals

Note: Existing performance data is reset (lost) when you change the interval configuration. However, only the data for that interval is reset. For example, if you change only the weekly time interval, the daily and monthly data are maintained,.

To configure performance intervals:

  1. Start VirtualCenter.
  2. Open the VirtualCenter Settings dialog and select the Performance tab.

  3. Modify intervals and sampling values as needed. Click OK.

    Name — the name of the interval. This name is selectable in the Ranges menu of the Performance tab.

    Minutes per sample — the length of time between each sample taken.

    Number of samples — the maximum number of samples that are displayed in the Performance tab. As new samples are collected, old samples beyond the maximum number are purge from the Performance display, but are kept in the database for historical reference.

    The charts in the Performance tabs reflects the adjusted values.

Configuring the Number of Threads for Collecting Performance Statistics

Configuring the Number of Threads for Collecting Performance Statistics

You can configure multiple statistical refresh threads and distribute their updates over time. On an SMP VirtualCenter host, use multiple threads to increase the number of managed hosts and take advantage of the additional CPU. On a uniprocessor host adjust this value to smooth out CPU usage.

The number of threads is used for collecting performance statistics from managed hosts. The default value is 1, the maximum setting is 8.

To configure number of threads:

  1. Start VirtualCenter.
  2. Open the VirtualCenter Settings dialog and select File > Advanced menu option.
  3. Select the Performance tab in the VirtualCenter Settings dialog box.

  4. Modify perf.numThreads value as needed. Click OK.

    On SMP hosts, using multiple threads increases the number of managed hosts possible by taking advantage of the additional CPUs. On UP hosts it improves the CPU utilization.

  5. Restart the VirtualCenter server to apply the changes.
Exporting Performance Data

Exporting Performance Data

Performance data can be exported to various file types.

To export performance data:

  1. Select File > Report > Performance and Report > Summary.
  2. Select a file format type to save out the data.

    File options include:

    • FormatedFiles.cs -- For large files dealing with the various file formats. Currently CSV, Excel/XLS, XML, HTML (with internal/external CSS support), TXT.
    • HostSummaryFile.cs -- File for controlling the layout and output of the host summaries accessible through the main form.
    • FormExcelOutput.cs -- A dialog box class for configuring various graph options for Excel output.
    • ExcelGraph.cs - For large files that deals with Querying the VMDB for graphing data and then generating an excel graph from it. Includes a Graph class for constructing/customizing graphs.
  3. For additional formats, select Save from the Export Event dialog.

previous Prev   Contents   Last   Next next