HP Capacity Advisor 7.3 Update 1 User Guide Abstract This guide describes how Capacity Advisor can help you plan for resource usage changes in a virtualized server environment, and analyze system data to flag any potential problems. Examples help you with common planning and analysis tasks. This document is for technical professionals working in data center operations, administration, and planning. Some understanding of system administration is assumed.
© Copyright 2006, 2014 Hewlett-Packard Development Company, L.P. Confidential computer software. Valid license from HP required for possession, use or copying. Consistent with FAR 12.211 and 12.212, Commercial Computer Software, Computer Software Documentation, and Technical Data for Commercial Items are licensed to the U.S. Government under vendor's standard commercial license. The information contained herein is subject to change without notice.
Contents 1 Introduction.............................................................................................11 Setup considerations...............................................................................................................12 Licensing requirements........................................................................................................12 Installation......................................................................................................................
With percentage of time limits....................................................................................30 Scope of utilization limits................................................................................................30 Adjusting power.....................................................................................................................31 Power cap data.................................................................................................................
Creating a population report..........................................................................................56 Creating a trend report..................................................................................................56 Creating a power report................................................................................................57 Setting utilization limits............................................................................................................
Assigning VM hosts to a VMware DRS cluster in a scenario.....................................................77 Initially converting VM hosts to simulate a VMware DRS cluster...........................................77 Adding VM hosts to a converted cluster............................................................................78 Removing a VM host from a VMware DRS cluster..............................................................78 Removing a VMware DRS cluster from a scenario.....................
Readiness information......................................................................................................102 Working with the Analysis tab...........................................................................................103 Analysis tab menu bar.................................................................................................103 Server and data range selector.....................................................................................
Task: Understand current resource usage..................................................................................135 Task: Plan server consolidation...............................................................................................135 Understanding the consolidation task.................................................................................135 Example consolidation: Stacking applications on an existing server...................................
HP Smart Solver error messages.............................................................................................180 When available systems lack sufficient capacity...................................................................180 When the threshold for invalid data is exceeded..................................................................181 HP Smart Solver informational messages..................................................................................
capupgrade (1M).................................................................................................................214 Units and terminology................................................................................217 Snapshots of Capacity Advisor reports.........................................................221 Consolidation candidates report.............................................................................................221 Scenario comparison report......................
1 Introduction This document describes how to use HP Capacity Advisor to plan for resource usage changes in a virtualized server environment, and analyze system data to flag any potential problems. Examples help you with common planning and analysis tasks. This document is for technical professionals working in data center operations, administration, and planning. Some understanding of system administration is assumed.
• Number of CPU cores • CPU benchmark data • Memory size • Network I/O bandwidth • Disk I/O bandwidth • Power usage • Platform multiplier for memory (see “Adjusting for platform changes” (page 187)) • Virtualization adjustments (see “Adjusting for virtualization changes” (page 191)) Setup considerations To prepare to use Capacity Advisor, consider the following: • Licensing requirements • Installation • Upgrade and reinstallation • Credentials • Dependencies Licensing requirements
• On a Microsoft Windows CMS: Allot 64.5 MB for each workload on each Windows or Linux managed node to support agentless data collection. Allot 34.7 MB for each workload on each HP-UX managed system. • On an HP-UX CMS: Allot 93.1 MB for each workload on each Integrity Linux managed node to support agentless data collection. Allot 34.7 MB for each workload on each Integrity Windows or HP-UX managed system.
different view. This can cause problems, including incorrectly displaying which objects are selected in the current view, losing information input into screen fields, and cancelling operations. Instead, use the navigation links, buttons, and tabs provided by the user interfaces within the browser screen area.
2 Features Capacity Advisor lets you test configuration changes before actually making them in your data center, and helps you to manage existing resources to improve their utilization.
• ◦ Import existing workload profiles ◦ Edit workload demands ◦ Move workloads between systems ◦ Park workloads ◦ Delete workloads For scenarios themselves: ◦ View and undo/redo scenario changes ◦ Use the HP Smart Solver technology to perform: ◦ – Automated server consolidation to virtual machines – Automated load balancing of existing VM hosts or existing servers – Automated workload stacking (consolidation) Create scenario comparison reports Data collection Utilization data collect
Figure 1 Capacity Advisor data collection infrastructure 1 2 3 4 5 6 The Agentless Data Collector Services run on the Microsoft Windows CMS and collect data from Windows managed systems. Agentless data collected by the Agentless Data Collector Service is made available to the Capacity Advisor Data Collector. The Utilization Provider enables the operation of the Capacity Advisor Data Collector. HP Insight Control virtual machine management data is made available to the Capacity Advisor Data Collector.
Collecting data for a workload with a HP Serviceguard package You can associate a monitored workload with a Serviceguard package. This association allows Matrix OE visualization to show workloads moving from one system to another within minutes of a Serviceguard migration. It also allows Capacity Advisor to show an uninterrupted history of utilization data for a workload across migrations.
Table 2 A comparison of agentless data collection and use of Utilization WBEM Provider Agentless data collection Utilization Provider Equally useful in data quality and accuracy. Available for Microsoft Windows, Linux, and non-HP systems Available for HP-UX and OpenVMS systems Provides the basic set of utilization metrics for CPU, Provides the basic set of resource utilization metrics for memory, network I/O, and disk I/O, plus additional CPU, memory, network I/O, and disk I/O.
Differences in network data ICperf does not report the correct network I/O when network interface cards (NICs) are being teamed for network fault tolerance (for example, by using HP Network Config utility). That is, for a system with two NICs teamed as one single logical NIC, Utilization Provider reports the addition of TX and RX Mbps for the teamed NIC. ICperf adds the metrics from the two NICs, ignoring that they are working as a teamed NIC.
Figure 3 CPU utilization for managed system puny03v7 1 Peak value. Comparing these two graphs shows that workload peaks on the two systems do not occur simultaneously, nor do they require the same percentage of the allocated CPU cores for processing.
Figure 5 CPU utilization with selected percentiles of utilization In this example, an allocation of three CPU cores is assumed, rather than the four cores initially available in this exercise. Even this reduction may not provide the best fit, as the majority of the work (90%) is completed with less than .5 of one core, and 99% of the work is completed with one core. For example, as shown in Figure 5 (page 22), CPU utilization has one peak at 1.7 CPU cores, with many lower peaks.
3 Key Capacity Advisor concepts Capacity planning goals Capacity planners are driven by multiple, sometimes conflicting, goals: • Avoiding surprises • Planning for the future • Maintaining quality of service (see “Quality of service ” (page 23)) • Optimizing resource utilization • Meeting internal and external security requirements • Reducing costs Capacity Advisor can help lessen the conflicts among goals by buttressing decisions with better information and more robust models.
• Capacity Advisor usually reports CPU utilization that is greater than what is reported by the VM guest. The operating system inside a virtual machine only knows about the CPU time used by threads that simulate the virtual processors. However, there are also threads that simulate the I/O cards and disks in the virtual system.
choose, select Configure→Edit Network and Disk I/O Capacity... from the Visualization tab in Matrix Operating Environment. Disk I/O The maximum capacity for disk I/O defaults to the high-water mark (that is, the maximum observed or collected value) for disk I/O. However, to set the upper bound to a value you choose, select Configure→Edit Network and Disk I/O Capacity... from the Visualization tab in Matrix Operating Environment.
you have a system with four cores where you never want utilization to exceed 75%, and peak utilization is 1.75 cores, then headroom is 1.25 cores. Optimum headroom varies depending on size of system. While a single processor system might require 50% headroom to preserve reasonable response times, a 16-way system might have reasonable response times when loaded at 80%.
(increased by 100%, or twice the current size) from their utilization during the selected reporting period. Large values for relative headroom The scenario comparison report provides indicators of relative headroom for CPU and memory use for two – four selected scenarios.
NOTE: In some situations, where time or time zones on a server are incorrect, it may appear that only old data is available in the data collection. For more information on this topic, see the section “Handling Old Data” in Chapter 9 (page 177). The affect of invalid data on HP Smart Solver solution finding The HP Smart Solver is set to ignore workloads having >25% invalid data; that is, the Solver will not consider these workloads within an automated solution.
TIP: You can specify more than one utilization limit for a resource Using the Utilization Limits Editor, you can add multiple settings for a resource. For example, you can create multiple different utilization limits for CPU cores by varying percentage and allowed duration for each limit.
Table 3 Percent of time conversions (continued) Percent of Time Minutes/ Week Hours/ Week Hours/Day (24–hour day) 30 3,024.0 50.4 7.20 100 10080.0 168.00 24.00 Understanding utilization limit messages Percentage of allocation The utilization limit messages are shown as a percentage of allocation, where allocation is subset of the given hardware for the system the workload is running on. For example, for a one-core system, the allocation is one CPU.
Table 4 Scope of utilization limits Scope Limits More global Global Utilization Limit Description Overrides • Applies to all workloads for which a more specific utilization • Nothing limit is not provided. • Cannot be disabled Workload Utilization Limit • Applies to a specific workload unless a more specific utilization limit is provided.
Capacity Advisor gives you insight into the current power cap for HP systems that have these options available and operating. Within Capacity Advisor, you can generate a power report to see historic power usage data that includes power cap settings and estimated enforcement time. The following image shows an example of the power cap data included in a Capacity Advisor power report. The power caps settings shown are provided by the administrator who is managing power consumption of servers.
• Trends are frequently small values, on the order of percents or fractions of a percent per month. • The cyclic data can easily be orders of magnitude greater than the trend (heavy calculations the day before payroll distribution, floods of users logging on after work on the East coast, and so on). • Special events can also be orders of magnitude greater than the trend (seasonal promotions, once per year calculations such as taxes). Any algorithmic analysis must be able to deal with these problems.
The Validity Threshold that you set should reflect your tolerance for obtaining a sufficient amount of valid data in the collection period that you designate. If the reports that you run show that the given threshold is not obtainable for the designated time period, this may indicate that many of the data points in the designated collection period are invalid.
Table 5 Forecast models Forecast Description Overrides Global Forecast Model Applies to all workloads in Capacity Advisor for which a more specific forecast is not provided. • Nothing Workload Forecast Model Applies to a specific workload in Capacity Advisor unless a more specific forecast is provided. • Global Scenario-wide Forecast Model Applies to all workloads within a Capacity Advisor scenario for which a more specific forecast is not provided.
NOTE: Combining the data range with the annual growth rate The forecast is applied point-by-point to the historical data within the range specified by the user. It is applied linearly, so that a point one year from the starting point of a forecast is the result of the full growth rate being applied to the data.
4 Procedures This chapter provides information on procedures you are likely to use with Capacity Advisor. Accessing Capacity Advisor Before you can use Capacity Advisor, you must access it. This procedure assumes the following: • You are familiar with Systems Insight Manager. For more information, select Help→For HP Systems Insight Manager from the top menu bar.
Data retention period By default, data is kept on the CMS for a period of four years. The amount of time can be altered so that it is shorter or longer than the default period. There is no upper limit on the number of years that data can be kept; however, you are restricted by the amount of disk space available. Procedure 2 Changing the data retention period 1. 2. Navigate to the vseprefs.
view in this same area, or by opening the Visualization tab (Tools→HP Matrix OE visualization... from the top menu bar). You may also have servers in mind where data collection has been accomplished using software other than the Utilization Provider. This can be done by collecting data from licensed systems for which Capacity Advisor has agentless data collection capability or by importing data collected by other agents, such as HP Performance Agent (OVPA) or Insight Control performance management.
Procedure 4 Windows CMS: Starting the HP Capacity Advisor Data Collection Service 1. 2. Select Start→Administrative Tools→Services. In the Name column, look for “HP Capacity Advisor Data Service”. Right-click on this name and select Start from the menu. Procedure 5 HP-UX CMS: Stopping the HP Capacity Advisor Data Collection Service • From the command line as root, run /sbin/init.
• From the Matrix OE Visualization tab or the Planning tab, select Configure→Global Utilization Limits to set default limits for utilization of CPU, memory, and disk and network I/O bandwidth to better reflect your Service Level Agreement. See “Setting global utilization limits” (page 58). • Determine your expected growth rate, then use Configure→Global Forecast...to realistically project future utilization.
3. Optionally, change the polling interval. It is unlikely that you will need to change this interval: 4. 5. 6. • Changing the data collection interval does not change how much data is stored. If data is collected more often, the multiple readings within each five minute interval are averaged together and stored as a single reading. • Because Windows now uses a 64-bit counter for accumulating network traffic, you do not need to sample more often to prevent the counter from overflowing.
Working with Capacity Analysis The Capacity Advisor Consolidation software does not work with Capacity Analysis. Systems with this license are not included in the analysis performed by Capacity Analysis and do not appear in analysis query results. Importing data for use in Capacity Advisor To import system utilization data, the system must be discovered by HP SIM.
5. 6. Optionally, specify [-p] if the import is for a non-Matrix OE workload. Click Run Now. After a short time, the Task Results screen displays. For tasks where you select Run Now, a few moments will pass and then you see the Task Results screen. For general help on interpreting the information in this screen, see the HP Systems Insight Manager Help on “Viewing Task Results”. The Task Results screen shows a “Running” check box with a Start time: until the task completes.
the first argument. (See the corresponding command reference pages in Appendix (page 197) for more information.) For example: # capovpaextract node12.company.com Ascertaining the data collection availability for a set of servers From within a scenario, or from the Matrix OE Visualization tab, you can see the data collection period for systems included in a scenario or for individual managed systems, respectively.
When a virtual machine is running HP-UX 11.23, most built-in commands such as top and sar will not be aware of the dynamic memory feature and can show memory sizes that are different from what is recorded by Capacity Advisor. For example, if a virtual machine is initially booted with 16 GB of memory, and is then re-sized to 4 GB, the built-in commands will not know about the change and will show 16 GB for physical RAM. Capacity Advisor, however, will show that the available memory for the VM is 4 GB.
Prerequisites • You must be logged in to Matrix Operating Environment (see Accessing Capacity Advisor).(see “Accessing Capacity Advisor” (page 37)). • You must have collected data on the systems of interest (see “Gathering data for Capacity Advisor” (page 37)). Some functions of a profile viewer require Capacity Advisor Tools permissions to be seen and used.
Procedure 12 To access a profile viewer when editing a scenario This procedure gives you access to view simulated resource utilization on a simulated system. This procedure assumes that you are starting from the Capacity Advisor Edit Scenario screen in the System tab view. For information on getting to this location, see “Working with systems” (page 73). 1.
3. Click OK to save the changes for each setting that you edit, and return to the Profile Viewer. NOTE: Viewing a VM guest on a VM host or viewing a complex In certain situations (such as when you are viewing the profile for a system or complex), navigational controls are presented to the right of the Hierarchy label near the top of the screen, such as links and drop-down lists , that allow you to view parents and siblings of the current resource profile.
• Click the radio button to the left of the metric that you want to view. The user interface pauses momentarily to reset the screen display to the new metric, updating both the utilization graphic and the Interval Metric Summary table. Using the report wizard Capacity Advisor reports provide valuable summaries and details from data collected from managed systems.
1. Click one of the visible links to select a scenario, one or more collections, one or more complexes, one or more systems, or one or more workloads to appear in the report. (The availability of selections depends on the report type selected.) A new screen opens with an appropriate list of known scenarios, collections, complexes, systems, or workloads. 2. 3. Click the check box to the left of each object's name that is to appear in the report. Click OK.
4. Select details specific to report type. a. Select one or more types of resources for which you want to see data from among CPU, memory, network I/O, disk I/O, and power. b. Select the type of scale for data presentation: percentage of allocated resource, absolute numbers, or both. c. Check the default graph size. The default size fits just within a letter-size page for printing. If you prefer dimensions smaller or larger, use the drop-down to change width (in pixels).
• Estimating the effect of moving processors (page 151) • Determining where to put a workload using automated solution finding (page 152) Using the report wizard to create a scenario comparison The Scenario Comparison Report allows you to compare two to four scenarios in a tabular and graphical side-by-side comparison report. Procedure 20 Selecting scenarios and setting the data range 1. 2. From the Planning tab, select Report→Create Scenario Comparison Reports... .
Figure 6 Comparison of three scenarios in a Capacity Advisor report This report includes two consolidation solutions provided by the Smart Solver (ServerConsolidationResult-SmallSystem and ServerConsolidationResults-BigSystem) and the original pre-consolidation set of physical servers (ServerConsolidationScenario). For more snapshots from this same report, see “Scenario comparison report” (page 221).
The ratio for the small system consolidation is 20/6, and the ratio for the big system consolidation is 20/3 – meaning that the 20 physical servers can fit on six small VM hosts or three larger VM hosts. Creating a consolidation candidates report 1. 2. Select Reports→Capacity Planning→Create Consolidation Candidates Reports... The Report Wizard opens on the Select Report Targets screen. Select the collection, complex, scenario, or systems for which you want the report, and select the data range.
Creating a peak summary report 1. 2. Select Reports→Capacity Planning→Create Peak Summary Reports.... The Report Wizard opens on the Select Report Targets screen. Select the systems, scenario, or collection for which you want the report, and select the data range. See Select targets and set date range for more information. 3. Select details specific to report type. 1. Select one or more types of resources for which you want to see data from among CPU, memory, network I/O, and disk I/O. 2.
4. Select details specific to report type. a. Select one or more types of resources for which you want to see data from among CPU, memory, network I/O, and disk I/O. (Power is not available when reporting trends only.) b. Check the default graph size. The default size fits just within a letter-size page for printing. If you prefer dimensions smaller or larger, use the drop-down to change width (in pixels). If you prefer a different width-to-height proportion, use the Aspect Ratio drop down.
for a workload, a visual indicator along with a warning message will be displayed when moving a workload in a scenario. When making automated changes, such as the automated system consolidation, these utilization limits are enforced in determining a solution. For more discussion about utilization limits, see “Utilization limits ” (page 28). IMPORTANT: For specific descriptions of each field or summary table on the user interface screens, click the help topic link on the software screen for the task.
Setting workload utilization limits Utilization limits allow you to set service level objectives for workloads based on overall system utilization. For any given workload, you can specify one or more service level objectives. The limit acts as an indicator for what the desired maximum usage would be. This limit can be applied to selected workloads on existing systems viewable from the Matrix OE Visualization tab. The workload utilization limits apply only to the selected workload(s).
4. To see the current utilization settings, click Revert. To cancel your changes, click Cancel. To accept your changes, click OK. Enabling or disabling the limit To enable or disable the application of the current utilization limit on the scenario: • Select either Enabled or Disabled. Removing a utilization limit 1. For each utilization limit you wish to delete: a. Check the box next to the utilization limit. b. Click Remove. 2. To see the current global utilization settings, click Revert.
To return to a planning checklist: • Consolidating server loads onto a virtual machine manually (page 136) Forecasting utilization Capacity Advisor provides multiple levels of forecast models to allow precise control of forecasting future utilization. IMPORTANT: For specific descriptions of each field or summary table on the user interface screens, click the help topic link on the software screen for the task.
2. Choose a time frame to use in tiling the data into the future (default: fixed interval). The time interval (date range) field adjusts according to your time frame selection. 3. Select the date range for defining the period of data collection to use as the basis for the forecast. a. For a fixed interval: select the calendar interval, Beginning or Ending, and the date (MM-DD-YYYY) b. For a ranged interval: select the beginning and ending dates and times. c.
1. If you are not on the Visualization screen, from the top menu, select Tools→HP Matrix OE visualization.... The Visualization tab screen is displayed. 2. • To define the forecast model for a system, start on the Visualization tab. • To define the forecast model for a workload, click the Workload tab. The Workload tab opens, displaying the licensed workloads. 3.
Prerequisites • 1. You must be editing a scenario (see “Editing a scenario” (page 67)). If you are not viewing the workloads, click the Workload tab. The Workload tab opens, revealing a list of the workloads in the scenario. 2. 3. Click the check box preceding a single workload. Select Edit→Scenario Workload Forecast... from the menu bar. The Forecast Editor: Scenario Workload Forecast Model screen is displayed. 4. • To define the forecast model, follow the procedure Defining a forecast model.
Prerequisites: • • You must be in a profile viewer. (See “Using the Profile Viewer” (page 46).) Change the Time Frame and/or Interval and/or Beginning and/or Ending for the profile to include future dates and/or times. If necessary, click the Refresh button. The graph will be updated. Historical utilization data will be plotted on a white background and forecast data will be plotted on a light blue background.
TIP: As the number of scenarios grows, the description field can help you to differentiate similarly named scenarios. 4. Optional. You can de-select the check box for opening a scenario editing session after completing scenario creation. When you un-check this box, you will return to the scenario list page upon completing the scenario creation.
• Consolidating server loads onto a virtual machine manually (page 136) • Consolidating server loads onto a virtual machine using automated solution finding (page 147) • Estimating the effect of adding processors (page 151) • Estimating the effect of moving processors (page 151) • Determining where to put a workload using automated solution finding (page 152) • Estimating potential power savings across a range of servers (page 153) Editing a scenario Editing a scenario allows you to modify the s
Prerequisites: • You must be editing a scenario (see “Editing a scenario” (page 67)) NOTE: 1. Changes to the data collection period persist across sessions. Click Edit Interval within the Set Data Range box. The fields are enabled as shown in the following image. 2. Click the first drop-down arrow to the left to see the choices for the duration of the data collection period that you want to use. Choose one to view in the display. The choices are as follow: 3.
• Click the drop-down arrow to select a type of calculation to be represented in the meter bars in the resource utilization table. The choices are as follow: • average • 90th percentile • peak (default) • max 15 minute sustained The scenario resource utilization table displays the new values for each resource, including violations of utilization limits set for each resource.
• If the Planning tab is visible, click the Planning tab. • If the Planning tab is not visible, select Optimize→Capacity Planning→View Planning Scenarios... from the top menu bar. Procedure 32 To rename a scenario The Planning tab opens with a list of existing scenarios that you are authorized to view. 1. 2. Click the check box preceding the scenario that you want to rename. Select Modify→Rename Planning Scenario... from the menu bar. The Rename Scenario screen displays. 3. 4.
Getting there If you are not on the Planning tab: • If the Planning tab is visible, click the Planning tab. • If the Planning tab is not visible, select Optimize→Capacity Planning→View Planning Scenarios... from the top menu bar. From there: 1. Click the name of the scenario that you want to review. The scenario editor window opens in a separate browser. 2. Select What-If Action→Undo/Edit/View Applied What-If Actions from either the System or Workload tab.
To edit applied what-if actions If you wish to disable changes, click the check boxes preceding the changes you want to disable. You can check either one step or a range of steps that have been grouped together. Then, click Disable. If you want to re-enable changes, click in the check boxes preceding the changes you want to re-enable. You can check either one step or a range of steps that have been grouped together. Then, click Enable.
3. Select Delete→Planning Scenario from the Planning tab menu bar. The Delete Scenario confirmation screen will be displayed, with the name of the scenario that will be deleted. 4. Confirm that this is the scenario you want to delete, then click the OK button. You will be returned to the Planning tab screen. Working with systems Systems can represent physical hardware, Virtual Machines (VM), or Virtual Machine Hosts.
b. Customize system (using attributes from an existing system) 1. Sort or filter the list of existing servers to find the server you are most interested in modeling. 2. Select the desired server. 3. Review the automatically populated system attributes that appear below the system table. Fill in any fields that are empty for more accuracy in the simulation. 4. 5. 6. Optional. Alter the attributes to make the “what-if” system(s) more representative of the system that you want to simulate. c.
To return to a planning checklist: • Consolidating server loads onto a virtual machine manually (page 136) Editing a system When constructing a scenario, it is frequently necessary to modify the characteristics of a system to reflect future changes to existing hardware or to reflect the characteristics of a planned system. 1. If you are not on the System tab within the Edit Scenario window, click the System tab. The System tab opens, displaying the list of systems currently in the scenario. 2. 3.
TIP: A longer data collection period provides a better picture of activity on the system over time than shorter periods do. 5. 6. Change the Meters Represent setting if a different calculation is desired . Examine the table titled Systems to be modified:. If you wish to modify the CPU utilization to account for any change in virtualization overhead, you can enter a percentage under CPU virtualization overhead %. The default is 0% to represent no change. 7. Click OK to accept these changes.
8. Within the table titled To: (Selected VM Host), select a destination VM host for the previously selected systems. Server presentation: The table entries are ordered by headroom fitness as determined by Capacity Advisor, and Capacity Advisor selects the top selection by default. Examine the headroom rating and the utilization meters for each prospective host to determine which host best fits your goals for headroom fitness and optimal utilization.
3. Type a unique name for identifying the cluster (required). IMPORTANT: While Capacity Advisor will inform you when the cluster name that you have designated is not unique (a cluster with the same name has been discovered by Systems Insight Manager), it cannot check names of clusters that have not yet been discovered or created. Use names that you believe will not be used in the future (during the useful life of the scenario). 4. 5. Optional. Select HA (High Availability) Cluster.
3. Close the Undo/Edit/View Applied What-If Actions window to return to the updated System tab. Moving a virtual machine Moving Virtual Machines (VMs) among VM hosts allows you to redistribute virtual equivalents of existing systems across physical servers. Capacity Advisor automatically takes into account use of dynamic memory by a virtual machine. NOTE: Performance for a virtual machine is usually calculated using the performance index calculated for its VM host.
9. Click OK to accept these changes. Unknowingly accepting unseen values: This can happen if you modify the headroom calculation after selecting the destination server, but then click OK without refreshing the destination table with the new values. The resulting change to the utilization values shown in the System tab view will differ from those seen in the previously shown destination table because the System tab will reflect the modified values.
Setting the upper bound values 1. Select to use the maximum observed network and disk I/O readings Or Manually specify the network and disk I/O capacity values a. b. Enter a value for network I/O in Mb/s. Enter a value for disk I/O in MB/s. NOTE: If desired, you can set the manual maximum value to be greater than the observed maximum. The designated values are used to populate the Selected Systems table and are applied elsewhere by Capacity Advisor. 2.
5. a. b. c. 6. If you intend to park the workload(s), confirm that the check box labeled Park/Unassign Workload is checked. Skip to step 7. If you previously selected a system, review the associated information given for that system. Then skip to step 7. If you want to select a system, uncheck the Park/Unassign Workload check box, if needed, to enable system selection. Select a system on which to run the workload(s). See Workload Location for more information. NOTE: 7. a. b.
4. 5. 6. If you are editing a workload that uses profile data, or you want to change the workload so that it will use profile data, modify those fields as appropriate. See “Determining estimated utilization assumptions for a workload” (page 190) for more information about the Copy Profile fields. If you are editing a workload that uses a static profile created for this scenario, or you want to change the workload that it will use static information, modify those fields as appropriate.
3. Select What-if Action→Move Workloads... from the menu bar. The Edit Scenario: Move Workloads screen displays. 4. 5. Change the Meter Style, if desired. Change the Data Range if a different size of data collection is desired, and click OK to refresh the table display. TIP: A longer data collection period provides a better picture of activity on the system over time than shorter periods do. 6. 7. Change the Meters Represent setting if a different calculation is desired .
Parking a workload When constructing a scenario, it can be desirable to temporarily remove the effects of workloads to reflect potential changes, without permanently removing the workloads. This is referred to as “parking” the workloads. 1. 2. If you are not on the Workload tab, click the Workload tab. Click the check boxes preceding the names of the workloads you wish to park. Check marks will appear in the check boxes and the rows containing the workloads will be highlighted, to indicate selection. 3.
IMPORTANT: For specific descriptions of each field or summary table on the user interface screens, click the help topic link on the software screen for the task. TIP: For lists of all menu options for Capacity Advisor features, see “Menus & tabs” in Capacity Advisor Help. Calibrating power on actual systems This screen allows you to directly calibrate the power settings for actual managed systems.
Procedure 37 Calibrating Insight Control power management • Click the “Calibrate” link. Results are posted in the Last Attempted Auto-Calibration table. If an error is indicated, read the “Troubleshooting” and “Notes” column for help on fixing the error. Procedure 38 Clearing current power management calibration values • Click the “Clear Calibration Data” link. Calibration values are deleted. Results are posted in the Last Attempted Auto-Calibration table.
Known problem: Collecting an HP iLO license: One of the steps in the checklist is to collect the HP iLO license. The following are known issues with this step: • • You must be able to ping the HP iLO function from the CMS using only the system name (not the fully qualified domain name). ◦ The Systems Insight Manager License Manager uses the system name, not the fully qualified domain name, regardless of system settings. ◦ Systems in the same DNS zone as the CMS will work fine.
To return to a planning checklist: • Estimating potential power savings across a range of servers (page 153) Calibrating power within a scenario This screen allows you to define the power calibration parameters for simulated systems in a scenario. These power settings specify whether or not to calibrate power, and if so, which settings to use for the calibration. This section describes the procedures for calibrating power for a single or for multiple systems within a planning scenario.
• On the System tab, click a Calibrate Power link in the “Power Utilization” column on the Edit Scenario system list table (This link appears when CPU information is available, but power has not yet been calibrated.) • Also on the System tab, you can select the check box to the left of the desired system, and select Edit→Calibrate Power (All Selected Systems) on the Capacity Advisor menu bar. • From the profile viewer for a specific system, click the Calibrate Power link.
TIP: For lists of all menu options for Capacity Advisor features, see “Menus & tabs” in Capacity Advisor Help. Automated solution finding: System consolidation to VMs This section describes the procedure for performing an automated system consolidation to virtual machines. NOTE: A consolidation adheres to the existing utilization limits. For information on utilization limits, see “Utilization limits ” (page 28).
5. Depending on your host destination type, you will need to do one or both of the following steps. • View the list of potential hosts included in the scenario in the Select the existing hardware... table. Check the box(es) for the system(s) to be used as the VM host(s). TIP: • 6. You may need to expand the gray bar to view the table entries. If you select a host template, a template editor will open. Fill in the fields as desired.
Insufficient resources on the VM host: Smart Solver informs you when the destination VM host has insufficient resources to host all of the consolidated systems. This can happen when the selected systems do not have enough resource capacity to place all of the workloads or when workloads cannot be placed on any of the selected systems. The Smart Solver will try to place as many workloads as will fit, giving you a partial solution.
Systems involved in load balancing When load balancing is performed, the loads are balanced only across the resulting systems in the solution. For example, if only VM hosts A and B are used (and VM host C is not), then load balancing is performed only across VM hosts A and B. VM host C is not included for the load balancing calculation. Headroom rating shows zero (0) stars The headroom rating shows zero stars even though it appears that there is sufficient room on the VM host for the workloads.
If you want to load-balance servers, select What-If Action→Automated Load Balance of Servers... Step 1 of 3: Define the solution constraints 1. 2. Change the Meter Style, if desired. Change the Data Range if a different interval is desired, and click OK to refresh the table display. TIP: Longer intervals provide a better picture of activity on the system over time than shorter intervals do. 3. 4. Change the Meter Represent setting if a different calculation is desired .
limits, and headroom, where the goal is to distribute workloads so that each system has comparable headroom and therefore, similar headroom rating. Resource capacity Workloads that specify utilization limits for a metric (for example, memory or disk I/O) can only be placed on resources that define a capacity for the corresponding metric.
Automated solution finding: Workload stacking Capacity Advisor can automate a workload stacking solution, where you set which servers (existing or newly created) you want the workloads to be on and set optimization parameters. Capacity Advisor then calculates and displays the best options for moving the workloads based upon the parameters that you provided. This allows you to see the predicted results of moving existing workloads onto selected physical systems.
5. Depending on your host destination type, you will need to do one or both of the following steps: • View the list of potential hosts included in the scenario in the Select the existing hardware... table. Check the box(es) for the system(s) to be used as the destination server(s). TIP: • 6. You may need to expand the gray bar to view the table entries. If you select to use a host template, a template editor will open. Fill in the fields as desired.
To return to a planning checklist: • Determining where to put a workload using automated solution finding (page 152) Results: Automated workload stacking Expected results The solutions show the target systems with the selected workloads placed to require as few of the target systems as possible. The target systems are either existing servers, “what-if” servers generated by template, or a combination, depending on what targets you selected.
5 Using Capacity Analysis Introduction Data center management involves thousands of physical servers and tens of thousands of virtual servers. When there is a problem, the IT department is expected to resolve these issues in a way that minimizes the impact on business. Many issues can be detected or diagnosed by examining utilization data. Capacity Analysis is an automated tool that combs through gigabytes of utilization data and helps you locate likely problems.
Accessing Capacity Analysis After logging in to Matrix OE, you can access Capacity Analysis in several ways: • Select Tools→Capacity Analysis... from the top menu bar. • Select Optimize→Capacity Analysis... from the top menu bar. • Select Tools→HP Matrix OE visualization from the top menu bar, and then click the Analysis tab. Exploring the Analysis tab The Analysis tab (the Analysis Dashboard), is your starting point for understanding resource utilization on licensed servers.
Working with the Analysis tab The Analysis tab is composed of the following features: 1. 2. 3. 4. 5. 6. “Menu bar” (page 103) “Server and data range selector” (page 104) “Analysis Queries table” (page 105) Query history graph “Analysis Query Results table” (page 107) “Analysis time stamp” (page 108) Analysis tab menu bar The menu bar is at the top of the Analysis tab. Some of the menu options described in the following table must be enabled by first selecting an analysis query.
Table 7 Menu options on the Analysis tab (continued) Menu selection Action summary Actions→Delete... Delete an analysis query that you have created. To activate this menu option, click the radio button corresponding to the query that you want to delete. You cannot delete a predefined query or a query owned by another user. See “Deleting an analysis query” (page 133) for more information. Help→Analysis tab screen Launch a help page describing the Analysis tab.
Analysis queries table The predefined analysis queries, and any queries you create, are listed in a table below the server and data range selector. Use the table to see the analysis queries that you own (private) or that are available to all (public). (All predefined queries are public.) Table 8 Analysis queries table Column heading Description [radio button] Select a particular query in order to use a menu action (such as Open) on that query.
Filtering by name • You can hold your mouse over the empty field to bring up a tool tip about the expected input for the filter. • Enter characters that match the analysis queries you want to view. The relevant queries are displayed. • Click Reset to return the filter to its default state. All queries for which you have permission are shown again. Filtering by system There is no filter indicator in the Systems column.
Filtering by visibility • Click All to remove the default selection, which shows your private queries and all public queries. • Select public or private query visibility. Selecting the private check box displays only your private queries. If you have no private queries, the table is empty. An administrator with full Capacity Advisor permissions can see all private queries. • Click Reset to return the filter to its default state.
The data that is shown in the results is determined by several factors: • The construction and filtering of query expressions. A query defines a set of systems with specific characteristics, then filters the results based on whether systems meet any or all of the defined characteristics. See “Constructing a filter” (page 114). • The construction of result table columns. A query also defines the information that is included in the results table columns.
Predefined analysis queries Capacity Advisor provides a set of predefined queries that cover some commonly desired perspectives on resource use for a group of servers in the data center. These queries are listed in the tables below. The predefined queries cannot be edited, but you can use any of them as the basis for a new query with a new name. Predefined queries cannot be deleted from Capacity Analysis.
Table 10 Predefined analysis queries: Finding overused systems Query name Description Systems exceeding their utilization limits Shows systems with a Five-Star rating equal to 0. These systems may not be meeting their Service Level Agreements. It could be that the systems need more resources. Edit the utilization limits in the Configure menu of the Workload tab. See “Headroom star rating” (page 26) for more information about the star ratings.
Controlling power costs Power costs continue to climb in the data center. You can determine the power used by each system, and the relative performance of the system, to identify systems that are the least energy efficient to use, and good candidates for an upgrade.
Creating the query To create a custom analysis query: 1. Do one of the following: • • If you are creating the query from an existing query: ◦ In the Analysis tab, click the radio button next to the existing query. ◦ In the tab menu, select Actions→Open. If you are creating an entirely new query: In the Analysis tab menu, select Actions→New. The Analysis Query Definition screen is displayed. 2. In the Analysis Query Name field, enter a name for the query. A value in this field is required.
7. Optional. Select the information to be displayed in the analysis results for the query. a. In Table Columns, click Choose…. You can select any number of attributes for inclusion in the results. In effect, you are creating a customized report. See “Attribute selection” (page 117) for information on the attributes you can select. b. Click OK. Columns for the attributes you selected are now visible in the right-hand side of the Analysis Query Definition screen.
Character support in Capacity Analysis queries Use only supported characters in analysis query names, descriptions, and the query filter fields used for string comparison.
4. Select the attribute to be used in the expression: a. In the first drop-down list on the line, click the arrow next to Avg CPU (%). The Attribute Selection dialog box is displayed. b. Select an attribute from any of the following categories: • System For information on the available choices, see “System Tab” (page 118). • CPU For information on the available choices, see “CPU Tab” (page 119). • Memory For information on the available choices, see “Memory Tab” (page 122).
Filter creation tips • Placing a match All inside another match All, or a match Any inside another match Any, will be ignored. Any empty parentheses will be simplified when the query is saved. • The level on which an expression, or sub-expression, resides is shown by the indent. This helps in determining which expressions are grouped together. • You have the ability to drag-and-drop expressions. Be aware that all sub-expressions will move with their parent expression.
5. Create the second expression: a. Click the arrow for the first drop-down list. The Attribute Selection dialog box appears, opened to the CPU tab. b. c. d. Under CPU Capacity, select Core Count, and click OK. Click the arrow for the second drop-down list and select <=. In the quantity selection box, press the up arrow twice to set it to 2 CORES. The nested expression is now complete.
System Tab Table 15 Attributes available for use in analysis queries: System Tab Section Attribute Name Description General System Information System Name The system's host name, IP address, or VM name as discovered by HP SIM. An HP SIM administrator can manually set this value by editing the system properties. Hostname The system's fully qualified domain name as discovered by HP SIM. IP The system's primary IP address as identified by HP SIM.
Table 15 Attributes available for use in analysis queries: System Tab (continued) Section Attribute Name Description System Health and Fitness Five-Star On a scale of 0–5, this value indicates the system's available headroom across all resources. In a system with a rating of 0, one or more resources do not fit in the system and the system's utilization limits have been exceeded. A system with a rating of 5 has more than twice the needed headroom across each of its resources.
Table 16 Attributes available for use in analysis queries: CPU Tab (continued) CPU Capacity Proc Speed (GHz) Processor speed: The clock frequency of the system's processor. Core Count The number of physical (not logical) processor cores that are active on the system. CPU Headroom (%) The headroom percentage indicates how much the demand for the CPU could grow before violating the system's CPU utilization limits. This metric is capped at 3100% to avoid dividing by zero or other small numbers.
Table 16 Attributes available for use in analysis queries: CPU Tab (continued) Infinite Loop Detection Hours One Core Utilized The “Hours One Core Utilized” attribute is the number of continuous hours of that CPU load that has been at or above a 1 core load. This time interval always closes the analysis window and its value is between 0 and 1000 hours. Even when the analysis window is less than 1000 hours wide, the value can be as high as 1000.
Memory Tab Table 17 Attributes available for use in analysis queries: Memory Tab Section Attribute Name Description Memory Utilization (%) Peak Memory (%) The percentage of available memory used by the systems, measured in GB. 15–Min Memory (%) 90%ile Memory (%) Memory Utilization (GB) For more information see “Units and terminology” (page 217). Avg Memory (%) NOTE: The memory attributes only measure memory that is used by the applications running on the system.
Network Tab Table 18 Attributes available for use in analysis queries: Network Tab Section Attribute Name Description Network I/O Utilization (%) Peak Network I/O (%) Network utilization, as a percentage, is computed by dividing the observed network use in megabits per second (Mb/s) by the estimated network capacity for the system.
Disk Tab Table 19 Attributes available for use in analysis queries: Disk Tab Section Attribute Name Description Disk I/O Utilization (%) Peak Disk I/O (%) Percentage of the available bandwidth being utilized, measured in megabytes per second (MB/s). 15–Min Disk I/O (%) 90%ile Disk I/O (%) For more information see “Units and terminology” (page 217).
Power Tab Table 20 Attributes available for use in analysis queries: Power Tab Section Attribute Name Description Power Utilization (Watts) Peak Power (Watts) The amount of power used by the systems. This value is computed from the power calibration and may not exactly match the exact power consumption. 15–Min Power (Watts) 90%ile Power (Watts) Average Power (Watts) System Performance per Watt Power Capacity Optimal Perf/Watt The Optimal Performance/Watt.
• Use less than 1.6 cores most of the time • Use less than 3 GB of RAM most of the time Using the Capacity Analysis query language, this query can be expressed as follows: Match ALL of: Type = MICROSOFT HYPERV VMGUEST Core count = 4 15-Min CPU (cores) <= 1.6 15-Min Memory (GB) <= 3 Once you locate the large VMs that can be resized, you want to produce information to justify why a large VM could be exchanged for a smaller one. To do this, you decide to add the following information: • The average loads.
2. 3. Select the check boxes for the following attributes: • System tab: Select System Name and OS. • CPU tab: Select 15-Min CPU (cores), 90%ile CPU (cores), and Avg CPU (cores). • Memory tab: Select 15-Min Memory (GB) and 90%ile Memory (GB). Click OK. The selected columns are now visible on the right-hand side of the Analysis Query Definition screen. 4. Optional. Reorder, resize, or change the sort order of the columns. Defining the filter 1. 2.
Revising the query Perhaps you start using ESX in the data center in addition to Hyper-V, and you want to expand the query to work with either type of virtual machine. In the revised query, the Type needs to match either VMWARE ESX VMGUEST or MICROSOFT HYPERV VMGUEST. This corresponds to the Match Any option and requires a new sub-expression to include both types of virtual machine. To revise the query, follow these steps: 1. In the first line, click the icon. The Match drop-down list is displayed. 2.
Many IT organizations use the power of two for the number of cores in their virtual machines. This means they have VMs with one, two, four, or eight cores. Because the overhead for having a few extra cores in a VM is small, it is reasonable to standardize the number of cores per VM to the power of two. In terms of savings on the VM host, taking the time to decide if seven cores will suffice instead of eight cores is generally not worth the effort.
2. In the Analysis Query Definition screen: a. Enter a query name of VMs with one too many cores. b. Select the Viewable by all check box if others will be using this query, or leave it unselected if you will be the only person using the query. c. Enter a description. For example: Locate VMs with at least two cores where the 15-minute sustained CPU load would be less than 80% if one core was removed. The goal is to find cores that could be removed from the VMs to save licensing costs.
5. Click the blank area in the second Match Any line and drag it above the first expression. Both expressions should now have the same indent: 6. Complete the first expression: a. b. In the first line, click the icon three times to create three Match drop-down lists. Confirm that All is selected in each list. The query filter editor should look like this: c. In each Match All line, click the icon twice.
d. Use the drop-down menus to complete each sub-expression. First sub-expression: 15-Min CPU (%) < 40 Core count >= 2 Second sub-expression: 15-Min CPU (%) < 53 Core count >= 3 Third sub-expression: 15-Min CPU (%) < 60 Core count >= 4 7. Complete the second expression: a. b. In the expression's Match Any line, click the icon three times.
Changing analysis queries Editing an existing analysis query Prerequisites • The query must be a custom query. You cannot edit a predefined query. • You must own the analysis query that you plan to edit or have administrative privilege with full Capacity Advisor permissions. Editing the query 1. 2. In the Analysis tab, click the radio button next to the query that you want to edit. In the tab menu, select Actions→Open.
6 Planning with Capacity Advisor Getting ready To get maximum value from the Capacity Advisor tools, it is important to: • Be familiar with the HP Systems Insight Manager framework • Be familiar with the basic operation of Capacity Advisor • Be familiar with HP Matrix OE visualization • Have a clear question you are trying to answer • Have plenty of utilization data collected for Capacity Advisor • Have appropriate access roles on the servers about which you are developing the plan • Understand
Prerequisites • You must have already collected data on the systems of interest (see “Gathering data for Capacity Advisor” (page 37)). • You should be familiar with Capacity Advisor operations (see “Procedures” (page 37)). • You must be logged in to Matrix Operating Environment (see “Accessing Capacity Advisor” (page 37)). For specific descriptions of each field shown on the user interface screens, click the on the screen.
Step 1: Determine which systems to consolidate From the top menu bar, select Reports→Capacity Planning→Create Consolidation Candidates Reports... On selecting the systems link, a screen opens listing all discovered servers for which data has been collected (both physical and virtual machines). For the purposes of this exercise, assume that 20 physical servers are selected as targets, and that you are most interested in viewing CPU and memory data for these servers.
NOTE: While the default setting for a scenario is to recommend at least seven days of data to use for analysis, one carefully selected day can be sufficient to simulate your desired conditions; or you may need 30 days or more to reflect your business cycle. For best results, obtain a data set that best reflects the business interval that you want to simulate. 5. 6. Check the Summary and verify that the scenario name and selected systems have been entered correctly. Click Finish.
Figure 8 Example of bar meters on a System tab in the scenario editor The weekly data is representative of recent utilization and can be calculated quickly, but doesn't always give a comprehensive picture. For a more comprehensive picture, look at a month of data. Change the data range to a month by clicking Edit Interval and selecting “Month” from the first drop-down list, and then OK. Once the screen has refreshed, new information about resource utilization is available.
over the legacy08 CPU utilization bar, you can see a pop-up message indicating the CPU utilization limit that is exceeded: More information about the utilization limits can be obtained by moving your mouse over the workload name (legacy08_wl), as in this image: Now look at a few profiles for CPU usage. Click the legacy08 CPU utilization bar.
Looking at the legacy08 graph, you can see that processing frequently requires 100% of available CPU in a month-long period. Also, it appears that the activity on this system exceeds 70% of the utilization resource for 15 sustained minutes, the utilization limit found earlier for this system. By studying the Interval Metric Summary table, you can see that 90% of the application processing was measured at using .47 core or less. Less that 10% of processing measured on this server required more than .5 core.
The application on legacy03 has a two-core processing capacity. A significant gap exists between Average use (.35 cores) and Peak use (1.63 cores), as can be seen in the Interval Metric Summary. Comparing CPU core usage at the 90th Percentile (90% of usage measures fall below this value – 0.69 cores) with the Peak usage (1.63 cores), you can see that almost one core is required to support 10 percent of CPU use on this server.
By studying this table and the key, you may have already concluded: • None of the servers meet an acceptable level of fitness as described by the headroom rating. • Two servers (tornado and orthus) would require additional CPU and additional memory to host all 10 of the legacy servers. • One server (vse02) might have enough memory, but would require more CPU. Also, the demands on network and disk I/O would exceed resource capacity.
This time, vse02 looks like the best VM host candidate in this scenario, as long as additional CPU are added to it to handle the additional processing load. Adding processor cores A look at the peak usage data shows that the usage of the CPU core allocation varies among the five servers, but usage still appears to be within the bounds of a server having no more than eight cores. However, Capacity Advisor will expect to calculate CPU processing overhead for running each of the virtual machines.
Leave the default setting for headroom calculations as it is — exclude guest fitness results (no fitness rating will be considered for individual VMs in the calculation, only for the VM hosts as a whole). Study the To: (Selected VM Host) table and make adjustments if needed. For purposes of this example, the utilization limit will be changed. Click OK to complete the conversion of servers to virtual machines.
Click OK to save and apply the changes. Related topics • “Utilization limits ” (page 28) • “Setting scenario-wide utilization limits” (page 59) Automating the consolidation task The preceding sections demonstrated a basic progression through Capacity Advisor screens where each step required manual evaluation and adjustment. The time this takes is not too laborious for a few machines, but you may need to evaluate possibilities among hundreds of machines.
Table 23 Checklist — Consolidating server loads onto a virtual machine using automated solution finding (HP Smart Solver) Task Related Procedure(s) Determine which systems to consolidate (both to and from). • “Creating a consolidation candidates report” (page 55) Create a planning scenario. • “Creating a planning scenario” (page 65) Run reports on the scenario systems.
Related topic • “Creating a planning scenario” (page 65) Step 3: Edit the scenario The systems included in the scenario are listed on the System tab of the scenario editor. The bar meters in the table show the peak resource utilization from data collected for the current week (the default setting).
Now, decide whether to let Capacity Advisor select the defaults for hypervisor overhead, or to provide your own values. Click “More Information” for guidance on calculating hypervisor overhead. Next, set the constraints for the solver: CPU virtualization is set to 20%, as in the manual consolidation example. The default settings for load balancing and maximum allowed invalid data are not changed. Click Next to proceed to the next step.
Based on the information given in the solution, you might choose to do the following: • Run the Solver again, and define the template host to have 5 GB of memory to allow for more headroom in the systems and to obtain a solution that includes legacy03. Rather than starting over, click Return to Step 1.
you can make more informed decisions about re-configuring your servers to improve the quality of service or maintain it with a more efficient configuration. Capacity Advisor allows you to size your system with more precision. This sizing is not based on a simple peak utilization value, but on knowing the answers to questions such as whether your system is more than 70% busy for an unacceptable percentage of the time.
Table 26 Checklist — Determining where to put a workload using automated solution finding (HP Smart Solver) Task Related Procedure(s) Determine which workload(s) to move. See also “Task: Understand current resource usage” (page 135). • “Creating a consolidation candidates report” (page 55) Create a planning scenario. • “Creating a planning scenario” (page 65) Edit the scenario: • Modify the workload(s) as desired. • “Editing a workload” (page 82) • Modify potential host systems as desired.
Task: Estimating potential power savings across a range of server models For specific descriptions of each field shown on the user interface screens, click the on the screen. Table 27 Checklist — Estimating potential power savings Task Related procedure(s) Collect data for a period of time that fully reflects your business cycle(s). • “Ascertaining the data collection availability for a set of servers” (page 45) Check power calibration on existing servers.
This example scenario contains four older HP servers showing high levels of CPU usage, shown in the following image. Two of these servers are currently exceeding their CPU utilization limits (red bar indicator). However, the other servers are being allowed to utilize 100% of their utilization limit, well beyond the recommended allotment for CPU utilization. Note that power is calibrated for these systems, and they are running at maximum power usage.
While values for the selected model and server series are supplied in drop-down selectors from which you can choose, you will need to supply realistic values for network and disk I/O capacity that reflect your environment. Lastly, you will need to supply realistic values for power consumption for the server model selected. You can use HP power calculators to help you determine what these values should be.
The following image shows the summary data for power provided on the index page of the utilization report: Clicking a system name on the index page opens a detailed report for an individual server.
Move workloads from the older servers to the newer server From the Systems tab of the Edit Scenario screen, select all the servers in the scenario. Then select What-If Action→Change Servers to VMs→Automated System Consolidation to VMs. (See Automated solution finding: System consolidation to VMs for procedural help on this option. For this example, the Smart Solver load balanced after consolidation and a Virtualization Overhead of 20% was assumed.
Compare power use between NewVMhost1 running four VMs (including m-teal2) and m-teal2 running on an older physical server. Comparing energy use and costs across all servers included in a power consolidation scenario can provide a realistic estimation of savings achievable by moving work to servers with newer power saving technology. TIP: Though not discussed here, this planning task also demonstrated potential efficiencies in CPU and memory use.
Data modeling for VSP vPars and VMs Most systems fit in the two models, everything is dedicated, or everything is shared except sometimes memory. Physical nPar Original vPar VM VSP vPar CPU Private Private Private Shared Private Memory Private Private Private Mixed Private NICs Private Private Private Shared Shared HBAs Private Private Private Shared Shared HP Integrity VSP (Virtualization Services Platform) vPars have dedicated CPUs and memory, but share IO bandwidth.
hosts (See table). In the Capacity Analysis dashboard the only VSP systems available are VM guest and vPars. The following table indicates the VSP systems availability for the Capacity Advisor Scenario Editor: Table 28 System Available to Capacity Advisor Scenario Editor VSP Hosts hosting ONLY VMs YES VMs Hosted by VSP Host NO1 VSP Hosts hosting VPARs NO VPARS Hosted by VSP 6.2 Host NO 1 VM guests are not individually available to the Scenario Editor.
7 Using Capacity Advisor with HP Serviceguard You are likely to use both Capacity Advisor and HP Serviceguard together in your data center. Serviceguard organizes systems or nodes into Serviceguard clusters, called SG Members in Capacity Advisor screens such as the scenario editor and profile viewers. In a Serviceguard environment, applications, services, and other entities are organized as packages that can move from one cluster node to another. TIP: In the HP SIM Version C.05.
This process affects only the system workload for the virtual machine. If monitored workloads are defined inside the virtual machine, their data is tracked correctly even when the virtual machine fails over to a new host. It is a good idea to create a workload inside any virtual machine that you intend to fail over using Serviceguard to ensure that the utilization data of the virtual machine is kept whole.
8 Troubleshooting in Capacity Advisor Behaviors that you might see when working within the Capacity Advisor user interface that may not have error or warning messages associated with them are explained in this appendix. When there is no connection Within an interdependent solution like Matrix Operating Environment, there are conditions that can create errors in connection that make it difficult or impossible for some functions to complete in Capacity Advisor.
collection only. Systems with the HP Insight Capacity Advisor Consolidation software license are not included in the analysis performed by Capacity Analysis. For information on licensing, see the “License management” chapter in HP Matrix Operating Environment Getting Started Guide and “Data collection and the HP Capacity Advisor Consolidation software” (page 42).
No network connectivity between the CMS and the managed node This is often a source of errors, once a system has been configured and data collection is initially working, but subsequently fails.
And it includes this description: Check the comment of this event to see why the Linux agentless data collection output is invalid. If the "Complete output" is empty, SSH command execution likely timed out. Verify SSHD is working correctly on this node. And the Comments field contains: Error parsing the output of the command "date +%s"; field "export: Command not found." was expected to be an integer. Complete output is: export: Command not found.
Data seems to be incorrect or lost VM guest resource utilization appears to exceed the resource utilization of its VM host When viewing historical resource utilization data in a profile viewer, it can appear that the utilization of a resource by a given VM guest exceeds the utilization of that resource by its VM host at a given instance of time. (For example, you might see the I/O usage occasionally exceed the VM host's high-water mark value (the dashed blue line in a profile viewer graph).
Capacity Advisor 6.x or later to improve the accuracy of the performance measure. See “Creating a workload” (page 81) and “Editing a workload” (page 82). CPU utilization seems high on managed nodes running Microsoft Windows 2003 Server Agentless data collection might not be accurate on hyper-threaded systems. Suggested action: Update Microsoft Windows 2003 Server so that it correctly reports the number of physical hyperthreading-enabled processors or the number of physical multi-core processors.
• A system is initially discovered by IP address ONLY (for example, in a non-DNS environment). Later the system is identified in HP SIM by its fully-qualified host name (when DNS is subsequently enabled in the environment). • Also in a non-DNS environment, a Windows system may be initially discovered by its NetBIOS name. If DNS is later enabled in the environment, that system may be subsequently identified by its fully-qualified host name.
If the workload is not referenced on the Workload tab, you can delete the workload with the following command from the CLI: gwlm delete --workload=workload_name This will remove the association between the workload and its collected data. The next scheduled data collection (usually done nightly) will delete any non-referenced profile data. Capacity Advisor should correctly recognize the new server association from this point on.
Verify that the HP Capacity Advisor Data Service has computed which collections the current user is authorized to view The HP Capacity Advisor Data Service might not yet have determined a list of collections that can be viewed by the user. If the user is authorized to view some systems, then the list of authorized collections should be available within 20 minutes.
Exporting four years or 3 MB of data into Capacity Advisor for profiling capprofile aborts with data sets of this size. Suggested action: Export multiple, smaller chunks of data. Working with reports This section discusses problems that you may encounter when creating or printing reports.
scenario is created with systems having imagined names. When this occurs, the scenario change record (Undo/Edit/View screen) indicates an error; the scenario is corrupted and cannot be repaired. Suggested action: Create a new scenario, replacing the no-longer-unique “what-if” systems with new, uniquely named systems. Delete the corrupted scenario. Note that this naming corruption can also occur when using “what-if” workloads or “what-if” VMware DRS clusters in a scenario.
• Automatic power calibration can be done on managed ProLiant Linux systems with no additional agents or providers. • Automatic power calibration is supported for Integrity Linux systems when, in addition to the agentless data collection, the Utilization Provider (UP) is configured on managed nodes running supported Linux distributions. However, to obtain the UP, you must install it as part of the Integrity Linux distribution or from a separate web download.
systems from the top menu bar or from the command line. You need to be a privileged administrator user or root to do this action. You may then want to click the "Refresh Data" link on the Matrix OE visualization Workload tab to ensure the information is also up-to-date for Capacity Advisor. If this problem persists, you may need to run vseassist on the VM host to determine why Systems Insight Manager is not correctly identifying the VM host and its constituents.
2. Modify nameofSytem.txt • 3. Open the file saved and edit the following fields: OS and OSName ◦ WINNT – Windows, Hyper-V ◦ LINUX - Linux ◦ HYPERVISOR ESX ◦ HPUX – HP-UX ◦ OPENVMS - OpenVMS Import profile.txt • Run the following commands: 1. cd C:\Program Files\HP\Virtual Server Environment\bin 2. capprofile -i -o -S nameofSystem < nameofSystem.txt 4. Delete System of HP SIM. 1. In the left column ( System and Event Collections) of HP SIM, select All System. 2. Select the system. 3.
9 Capacity Advisor error and warning messages Certain messages that you might see when working within the Capacity Advisor user interface or that are about data collected for use in Capacity Advisor are explained in this section. Messages appearing in the user interface that are self-explanatory or that are supplied by the Utilization Data Collection Events listing in Systems Insight Manager are not included here.
Command warning messages Within the messages sections, messages appear in alphabetical order. Messages that start with system_name are placed at “s” in a list. These messages alert you to specific conditions present in the user interface . Follow-up actions may or may not be necessary. For additional information, refer to “Command reference” (page 197) in this guide, as well as the HP Systems Insight Manager Technical Reference Guide available at http://h18013.www1.hp.
Table 30 Warning messages in Capacity Advisor (continued) Message Probable cause To correct No WBEM credentials are available for system "system_name”. Please verify that a valid WBEM account and password for this system are configured in HP SIM (Options->Protocol Settings->Global Protocol Settings). Either no WBEM credentials have been entered in HP SIM or HP SIM could not successfully use the credentials entered. The latter can happen when acimserver process is not running on the system.
Table 30 Warning messages in Capacity Advisor (continued) Message Probable cause To correct The system "system_name" has no workload defined. Make sure to select Tools→VSE Management...in HP SIM before running this command for the first time. For HPVM Guests, please be sure that the HPVM WBEM provider is properly configured. If problems persist, run vseassist to further diagnose errors.
Remember that the Smart Solver includes utilization limits set for the workload, system, or scenario-wide when calculating desired capacity for the solution. Therefore, one way to correct the resource insufficiency could be to raise the utilization limits for the specified metrics. Another way is to select or add additional systems having at least the minimum required capacity.
• Collect or import additional valid data to improve the data validity percentage. • Adjust the workload forecast model data range selection to exclude invalid periods. If these steps are insufficient or cannot be done, remove the workloads from the scenario that have higher percentages of invalid data, and run the Solver on the remaining workloads to obtain the best placement for those workloads.
10 Support and other resources Information to collect before contacting HP Be sure to have the following information available before you contact HP: • Software product name • Hardware product model number • Operating system type and version • Applicable error message • Third-party hardware or software • Technical support registration number (if applicable) How to contact HP Use the following methods to contact HP technical support: • In the United States, see the Customer Service / Contact HP
The service also provides access to software updates and reference manuals in electronic form as they are made available from HP. Customers who purchase an electronic license are eligible for electronic updates. With this service, Insight Management customers benefit from expedited problem resolution as well as proactive notification and delivery of software updates. For more information about this service, see the following website: http://www.hp.
The following book can be ordered from InformIT Network (http://www.informit.com/store/ product.aspx?isbn=0131855220): • The HP Virtual Server Environment: Making the Adaptive Enterprise Vision a Reality in Your Datacenter, by Dan Herington and Bryan Jacquot, Prentice Hall, 2006. Related technical papers • Hewlett-Packard Development Company, L.P.; 2011. Backing up and restoring HP Insight Software 7.0 Central Management Server (Windows) at http://www.hp.com/go/matrixoe/ docs.
Calculation assistance See “Related technical papers” (page 185) for additional information. Cost calculation When creating utilization reports, Capacity Advisor gives you the ability to provide a value for calculating the cost per kilowatt-hour for selected servers in relation to the resource usage on those servers. Cost per kilowatt-hour Meaning This field represents the effective cost per kilowatt-hour that is paid for powering a server in a data center.
• • A move from one system architecture to another system architecture can increase or decrease resource utilization. A move from a two-way to a one-way system can decrease resource utilization. • A change in the application can increase or decrease resource utilization. (For information on how utilization is calculated for each resource, see Appendix (page 217).
CPU workload multiplier Meaning The relative change in CPU utilization desired when sizing an existing workload to better simulate a new workload in a scenario. CPU utilization by the new workload that you are creating can be made smaller, the same, or larger than that of the workload chosen as the baseline value. TIP: You do not need to account for different CPU clock speeds in this multiplier. Capacity Advisor will do this automatically. Default The default value is 1.0 (no change).
Disk I/O workload multiplier Meaning The relative change in disk I/O utilization desired when sizing an existing workload to better simulate a new workload in a scenario. Disk I/O can be made smaller, the same, or larger than that available for the workload chosen as the baseline value. Default The default value is 1.0 (no change).
Table 32 Define static estimates — settings to guide Estimated Utilization assumptions for workload Area Attribute or action Description Static Profile CPU Core Utilization Fractional or whole number of cores assumed to be used by the new workload on the assigned system. Default: 0.0 Memory Utilization (GB) Memory assumed to be used by the new workload. Default: 0.0 Network I/O Utilization (Mb/s) Network bandwidth assumed to be used by the new workload. Default: 0.
CPU Virtualization Overhead % Meaning The percent change in CPU utilization due to the overhead (or the absence of overhead) incurred by running an application in a virtual machine. When using virtual machines, the CPUs on the VM host do some of the work normally done by I/O cards. This work is the virtualization overhead accrued by running an application in a virtual machine.
Example: For a host with 32 GB of physical memory, the Hypervisor Memory Overhead will be: 750 MB (.73 GB) + 7.5% of 31 GB = .73 GB + 2.24 GB = 2.97 GB Source: System Sizing Guidelines for Integrity Virtual Machines Deployment VMware ESX 3 To compute the memory overhead of the hypervisor, use the following formula: Total Physical Memory – (Total Physical Memory – 284 MB)/1.078 This formula is derived from a least squares fit of observed values in test systems running VMware ESX.
IMPORTANT: Values given by these power calculators are only estimates. Actual results can vary depending on application load, ambient temperature, and other factors. Determining idle/max values for non-Blade servers 1. Locate the appropriate power calculator for a given model number. a. Download the power calculators in spreadsheet form for older HP Proliant DL/ML/PL servers: http://h30099.www3.hp.com/configurator/powercalcs.asp. For newer Proliant systems...
5. 6. 7. 8. a. Configure a single bay with a blade matching your desired configuration. b. Specify utilization to 1%. c. Select “Update Calculation” and record the Total System Input Power as idleSingle. Calculate the max value for single blade: a. Configure a single bay with a blade matching your desired configuration. b. Specify utilization to 100%. c. Select “Update Calculation” and record the Total System Input Power Requirement as maxSingle. Calculate idle value for two blades: a.
Command reference This reference section contains detailed descriptions of the Capacity Advisor commands. Commands available on HP-UX and Microsoft Windows operating systems The command information included here is specific to HP-UX, though these commands are also available to run on Windows. TIP: For options and examples that are specific to using these commands on Microsoft Windows, see the “Command Reference” in Capacity Advisor Help.
capcustombenchmark(1M) NAME capcustombenchmark -- Configure Capacity Advisor benchmark CPU scaling to normalize collected data.
result: value +/- std_dev : type • value is the performance index result. • +/- std_dev is the standard deviation for the type. • type is the method used to compute the index and can be CLOCK_SPEED, AVERAGE, EXACT_MATCH, or NORMAL. CLOCK_SPEED is the CPU clock speed for the container. From the example: ◦ ◦ ◦ value is 1.0 +/- std_dev is +/- 1.0 type is CLOCK_SPEED processor: The processor information gathered from Systems Insight Manager.
capcustombenchmark (4) NAME capcustombenchmark -- Format of Capacity Advisor custom benchmark database files. DESCRIPTION Defines the file format to use when creating custom benchmark data to be used by Capacity Advisor for scaling of CPU utilization values when moving workloads and/or virtual machines.
capovpaextract (1M) NAME capovpaextract -- Export Performance Agent (OVPA) system data from the specified managed node and import the data into Capacity Advisor. SYNOPSIS Path on CMS: • /opt/vse/bin/ capovpaextract [ -b begin-time -e end-time capovpaextract -h ] [ -p ] managed_node DESCRIPTION capovpaextract exports HP Performance Agent system utilization information from the specified managed node and imports the information into Capacity Advisor.
HP-UX • C.03.35 or later SOLARIS • C.03.75 or later LINUX • C.04.00 or later EXAMPLES Extract Performance Agent data and import the data to an existing system, test.company.com. # capovpaextract test.company.com Extract Performance Agent data for a specific time interval. # capovpaextract -b 01/01/06 -e 06/31/06 test.company.com Extract Performance Agent data for a specific time interval and import it as a non-Matrix OE workload to test.company.com. # capovpaextract -b 01/01/06 -e 06/31/06 -p test.
capprofile (1M) NAME capprofile -- Import, export, display, invalidate and remove Capacity Advisor data collected for workloads or systems. SYNOPSIS Path on CMS: • /opt/vse/bin capprofile profileID capprofile delimiter ] capprofile ...
The following cases cause imported data to be ignored or converted and a warning message issued: • A sample line with missing or extra metrics, as compared to the labels in the file header; the sample line is ignored. • A sample with negative values for one or more metrics. Negative values are converted to “Not A Number” floating point values. • A sample with invalidly formatted floating point numbers for the value of a metric; these sample lines are ignored.
cannot, for example, mark invalid only the CPU utilization portion of a sample. Use -m i to mark data for a specified interval and profileID as invalid; use -m v to mark data as valid. -o -p -r profile_ID -S -t -x -y delimiter When used with the -i option, overwrites existing data where there is an overlap in the date range of the import data and the date range of the existing data.
EXAMPLES Export profile data from December 14, 2005 at midnight to midnight December 31, 2005 for the workload with profileID billing3 to the file billing3.txt in the /tmp directory. capprofile -x -b 20051215 -e 200512312359 billing3 > /tmp/billing3.txt Import profile data from the file /tmp/billing3.txt for the profile with profileID billing3. capprofile -i -o billing3 < /tmp/billing3.txt Mark a range of utilization data as invalid for the workload with profileID prod05_wkld.
capprofile (4) NAME capprofile -- Format of Capacity Advisor import and export files. DESCRIPTION Defines the file format used when importing and exporting data for Capacity Advisor. The data for the specified profile is imported and exported as a series of lines containing the following: • profile headers, including date and metric labels • utilization values The collection period is five minutes. The collection occurs at the end of the five-minute period. Each data line is called a sample.
metric: One or more of the following metrics: CPU_UTIL, MEM_UTIL, NET_UTIL, DISK_UTIL, CPU_ALLOC, MEM_ALLOC, PHYS_CPUS, PHYS_MEM, CPU_QUEUE, PAGES_PER_SEC, IOS_PER_SEC, PKT_PER_SEC, DISK_USED. The command only outputs a result when there is data for the given metric during the provided period of time. If the metric data covers some of the time range, but does not encompass all of the time range, the column is padded with NaN values.
#CPU:4 @ 1.866GHz #Memory:4093MB #OS:WINNT #Model:ProLiant DL380 G5 #ProcessorString:Intel(R) Xeon(TM) Processor 1.866 GHz (x86 Family 6 Model 15 Stepping 7) #ProcessorFamily:Intel Xeon #Version:A.03.00.00 YYYYMMDDhhmm,UTIS,CPU_UTIL,MEM_UTIL,NET_UTIL,DISK_UTIL,CPU_ALLOC,MEM_ALLOC,PHYS_CPUS,PHYS_MEM 200908241200,1251136800,0.34790,2.12109,0.04915,0.66662,4.00000,3.99707,4.00000,3.99707 200908241205,1251137100,1.42610,1.93066,0.02458,1.41005,4.00000,3.99707,4.00000,3.99707 200908241210,1251137400,0.20913,2.
capagentlesscfg NAME capagentlesscfg -- Set options for Matrix Operating Environment agentless data collection, and view a configuration file that contains a list of nodes from which data can be collected. Agentless data collection is available for a CMS running Microsoft Windows for managed servers running Windows.
For the listed Windows systems, specify a 10-minute interval between data collections using 20 collector threads. # capagentlesscfg -m600 -n20 List all nodes configured for agentless data collection by this CMS: # capagentlesscfg -l AUTHORS capagentlesscfg was developed by Hewlett-Packard Company.
cappmpextract NAME cappmpextract -- Export HP Insight Control performance management (ICperf) system data from the specified managed node and import the data into Capacity Advisor. List ICperf-managed nodes where data can be extracted from or imported into Capacity Advisor. HP Insight Control performance management (ICperf) was formerly known as HP Performance Management Pack (PMP).
systems to use in Capacity Advisor scenarios. Non-Matrix OE workloads are not associated with any licensed nodes and are not visible on the Matrix OE Workload tab. To import non-Matrix OE workload data, use the following GUI action sequence starting from the Capacity Advisor tab: Modify -> Edit Planning Scenario. This will open the scenario that you selected. Then click What-If Action -> Create Workloads from the Edit Scenario menu.
capupgrade (1M) NAME capupgrade -- Recompute CPU utilization data. SYNOPSIS Default path: • Windows CMS: C:\Program Files\HP\Virtual Server Environment\bin\ capupgrade • HP-UX CMS: /opt/vse/bin/ capupgrade DESCRIPTION capupgrade is used to recompute the CPU utilization data after an upgrade of Matrix OE. A new copy of the Performance Index Database, which holds server performance benchmarks, is installed whenever you upgrade Matrix OE.
For more information, see C:\Program Files\HP\Virtual Server Environment\logs\ capupgrade.log/var/opt/vse/logs/capupgrade.log The Performance Index upgrade can take up to 3 minutes. During this time, the HP Capacity Advisor Data Service will be offline and data collection will be stopped. The service will restart automatically at the end of the capupgrade process.
Units and terminology The units and terms listed in the following table are used within Capacity Advisor. For an expanded list of terminology used within Capacity Advisor and Matrix OE, see the “Glossary” at the end of this guide. Table 33 Units and Terminology Unit or Key Word Meaning clock speed Some newer processors are capable of short bursts of speed above the base speed of the processor. When Capacity Advisor shows the clock speed of a system, it reports the base speed.
Table 33 Units and Terminology (continued) disk I/O bandwidth utilization Measured in MB/s (10^6 bytes, megabytes per second). Each sample represents an average reading over the past five minutes.
Table 33 Units and Terminology (continued) memory utilization The amount of memory used in gigabytes (2^30 bytes). Each sample represents an actual reading at the time the sample was taken.
Snapshots of Capacity Advisor reports Consolidation candidates report The following image illustrates a portion of data shown in a consolidation candidates report when CPU and memory resource utilization data are selected for inclusion. You can also select to see data on network I/O and disk I/O in this report. Other data that would appear in this report: average paging (pages/second). Note that the data can be downloaded to a CSV file for import into a spreadsheet.
Workload Detail 222 Snapshots of Capacity Advisor reports
Scenario Inventory Scenario comparison report 223
Graphed data in Capacity Advisor reports To access Capacity Advisor reports, select Reports→Capacity Planning . Table 38 Data graphs available in Capacity Advisor reports.
Known problems in Capacity Advisor documentation This guide contains the most recent information available for Capacity Advisor. This section is here to help you to confirm when information in Capacity Advisor documentation is inaccurate and to point you to the most up-to-date information. Errors in Capacity Advisor online help and command reference pages The online help refers to Insight Control performance management (ICperf) by its old name, Performance Management Pack (PMP).
Documentation feedback HP is committed to providing documentation that meets your needs. To help us improve the documentation, send any errors, suggestions, or comments to Documentation Feedback (docsfeedback@hp.com). Include the document title and part number, version number, or the URL when submitting your feedback.
Glossary 90th percentile That utilization value in the selected time interval which 10% of the utilization values fall above, and 90% fall below or are equal to. activate When referring to a logical server, activate means to make a logical server definition available to be deployed into the computing environment. An active logical server is one that is currently operating within the computing environment.
command line interface See CLI. complex A complex includes one or more cabinets that are cabled together and all of the hardware resources that they contain. A complex has a single Service Processor. See also server, system. constraints Resource allocation restrictions imposed by either the customer (for example, workload placement restrictions), or the Matrix Operating Environment. See also policy. core The actual data-processing engine within a processor.
hardware catalog A store of data containing physical and operational attributes of current supported models of HP servers. This data is provided to help you define realistic “what-if” systems to use in Capacity Advisor planning scenarios. headroom In general, the amount of a computing resource that is available on a system after all requirements for applications on the system are accounted for. In Capacity Advisor, requirements for applications include the utilization limits set for each application.
invalid data In Capacity Advisor, data that could potentially skew reporting results and lead to inaccurate or misleading conclusions when capacity planning. Examples of events that Capacity Advisor can recognize as potential sources of invalid points include the following: • System downtime during a data collection period • Data manually marked by the user as invalid • Partial collection from a virtual machine or a VM host See also missing data, valid data.
associated with a system may have historical data associated with it from Capacity Advisor or Global Workload Manager (gWLM). As with any workload, the historical data will be lost if the workload is deleted. When migrating a workload from one system to another, it may be useful to park the workload (removing the association with the original system) until the new system becomes available. This preserves the historical data for the workload across the migration. partition 1.
server 1. Physical server: Hardware that can run one or more operating systems, including a partitionable complex. Also, hardware that can run an instance of the vPars monitor. Server hardware includes one or more cabinets containing all the available processing cores, memory, I/O, and power and cooling components. HP Integrity servers include two types of server hardware: standalone servers and cell-based servers. 2. Virtual server: A software-based virtual environment that can run an operating system.
unbound core For vPars versions prior to A.04xx, an unbound core is one that can be migrated between virtual partitions while those partitions are running. Unbound cores cannot handle I/O interrupts. Unbound cores are sometimes referred to as “floater processors.” The distinction between bound and unbound cores does not apply to vPars version 4.0 or later. utilization limits The limits set on the usage of system resources such as CPU, memory, or network I/O by an application.
Web-Based Enterprise Management See WBEM. what-if scenario A configuration of systems and workloads that is different from the current configuration. Capacity-planning simulations are run using what-if scenarios as experiments before making any actual configuration changes. wizard A sequential series of pages that transforms a complex task into simple steps and guides you though them. The wizard makes sure that you provide all of the required information and do not skip any steps.
Index A absolute interval, 49 agentless data collection for Capacity Advisor viewing configuration, 41 Analysis Dashboard, 102 analysis query constructing a filter, 114, 127, 130 creating custom queries, 111 custom query examples, 125 deleting an existing query, 133 editing an existing query, 133 predefined analysis queries, 109 query attribute selection, 117 using sub-expressions, 116 Analysis tab, 102, 103 menus, 103 annual growth rate, 35 and data range combined, 36 assistance, 183 automated consolidatio
as measured for disk I/O, 25 as measured for memory, 24 as measured for network I/O, 24 as measured for power usage, 25 availability for use in Capacity Analysis dashboard, 102 collect for Capacity Advisor, 37 collect in Serviceguard cluster, 162 considerations in measuring resource utilization, 25 conversion, 17 export, 197 gap, 162 historical, 17 ICperf, 197 import, 197 incorrect, 167 invalid, 27 invalidating, 197 linear regression to calculate trend line, 34 measurement interval, 25 missing, 27, 162 old,
capacity planning, 23 growth rate forecasting, 188 H headroom, 218 defined, 25 interpreting star rating, 26 stars definitions, 26 help obtaining, 183 historical data, 17 historical utilization, 15 host name, 161 HP technical support, 183 HP iLO collecting license for power calibration, 88 HP Performance Agent, importing data into Capacity Advisor , 39 HP SIM documentation, 184 HP Smart Solver, 32 HP Virtual Machine calculating Hypervisor Memory Overhead for Capacity Advisor, 192 hpvmmigrate command, 162 Hy
N network data, 20 network I/O bandwidth utlization, 219 edit capacity in Capacity Advisor, 80 network I/O capacity, 24 Network I/O Utilization, 190 Network I/O Workload Multiplier, 189, 190 default value, 189 example, 189 no data for managed node, 45 O Offset Hours, 190 old data, 166 old dates, 166 operations overview Capacity Advisor scenarios, 65 Capacity Advisor systems, 73 OTHER workload, 219 out-of-memory, 171 overview Capacity Advisor procedures, 37 OVPA data, 17 importing, 43 P peak data, 15 peak
change the meter style of a scenario display, 67 change time and data range in profile viewer, 49 change view of system hierarchy in profile viewer, 48 collect Capacity Advisor data, 37 controlling Capacity Advisor data presentation, 67 copying a scenario, 69 creating a consolidation candidates report, 55 creating a cost allocation report, 55 creating a peak summary report, 56 creating a planning scenario, 65 creating a population report, 56 creating a power report, 57 creating a scenario comparison report,
workload, 161 SG Member, 161 simulator, 23 software technical support, 183 update service, 183 stars meaning in Capacity Advisor, 26 Static Profile creating a baseline workload, 191 parameters for workload in Capacity Advisor scenario, 190 static profile providing estimates for, 191 sums and peaks, 25 support, 183 sustained minutes limit example, 30 system add existing to Capacity Advisor scenario, 74 create for Capacity Advisor, 73 edit for Capacity Advisor, 75 overprovisioning error, 29 remove from Capaci
V valid data factors affecting reporting in Capacity Advisor, 33 selecting appropriate business interval, 33 setting threshold values for Capacity Advisor reports, 33 view Capacity Advisor profile viewer, 46 virtual machine, 15, 162 calculating performance when running in a cluster, 39 move in Capacity Advisor, 79 VM guest viewing in profile viewer, 49 VMware ESX calculating Hypervisor Memory Overhead for Capacity Advisor, 193 VMware vSphere calculating Hypervisor Memory Overhead for Capacity Advisor, 193 v