HP ProLiant Essentials Performance Management Pack User Guide Part number: 306919-408 Fourteenth edition: July 2007
Legal notices © Copyright 2003-2007 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 About this guide Where to go for additional help ................................................................................................................ 6 HP Services and Technical Support........................................................................................................ 6 Website............................................................................................................................................. 7 Authorized resellers...........................
Data reporting ....................................................................................................................................... 38 Static analysis report ......................................................................................................................... 40 System summary report ...................................................................................................................... 41 CSV file generator................................................
Sample Static Analysis Report.................................................................................................................. 69 Appendix D Sample server availability report ..............................................................................................................
About this guide This user guide provides step-by-step instructions for installing and using HP ProLiant Essentials Performance Management Pack. Where to go for additional help For additional information about HP ProLiant Essentials Performance Management Pack, refer to: • http://www.hp.
• Software and Drivers download pages (http://www.hp.com/support)—These pages provide the latest software and drivers for your ProLiant products. • Management Security (http://www.hp.com/servers/manage/security)—HP is proactive in its approach to the quality and security of all its management software. Be sure to check this website often for the latest downloadable security updates. • Obtain the latest SmartStart (http://www.hp.
Using the Performance Management Pack interface This chapter describes the user interface and tools provided with HP ProLiant Essentials Performance Management Pack (PMP). The browser-based user interface facilitates the use of PMP without having a specific management console. PMP data can even be accessed from a remote client if the environment security settings permit. The user interface is standard HTML and JavaScript.
Server status applies only to servers currently being monitored. An Unknown icon for a specific server in the PF column of the Systems Insight Manager console indicates that performance information is not available for that server. Clicking the Unknown icon for a server displays detailed reasons why the information is not available. An Unknown icon might appear for any of the following reasons: • A server is not licensed. • A server is licensed but not marked for monitoring.
The interface is divided into three frames.
“Recommendation,” providing possible actions to relieve the bottleneck, and the “Configuration Issues,” providing details about any issues detected for the components. The server metrics and their definitions for data analyzed by PMP to determine the status of a particular component are provided in Appendix A of this guide. • Graph—Displays a graphical representation of the performance metrics associated with the selected component.
Reports PMP can generate reports illustrating the percentage of time a system was in a bottleneck state and the overall performance utilization for a system categorized by its components. A System Summary Report can be generated from data in the performance database in HTML format. CSV File Generator can generate comma-separated value (CSV) files for import into desktop analysis or reporting tools.
Offline Analysis Offline Analysis is used to view recorded data sessions from the database, enabling review of specific logged information sampled from the database. For information about using Offline Analysis to view recorded data sessions, refer to Chapter 2 of this guide. During the session, data is shown in two refresh modes. • Auto mode—Plays the data back at the selected playback rate. Select the Auto Refresh checkbox to enable auto mode. • Manual mode—Enables you to advance the samples manually.
Manual Log Purge Manual Log Purge is used to delete logged files from the PMP database that are no longer needed. For information about using this tool to delete logged files, refer to Chapter 2 in this guide.
Using PMP This chapter provides an overview of PMP functionality. A usage scenario demonstrates setup, administration, and monitoring of servers. The usage scenario provides examples of bottleneck conditions with and without a recommended hardware upgrade, as well as examples of static and offline analysis. Opening the Systems Insight Manager console The Systems Insight Manager console, displayed in an Internet Explorer browser window, shows performance status in the PF column.
Debugging a server bottleneck condition The following sections detail the appropriate actions to take when a bottleneck condition exists on a server. In this scenario, the server name is “pmpserver.” Select the server Click the major icon to display the PMP Online Analysis window for “pmpserver.” The screen displays the server node tree in the left configuration navigation pane and the Status tab in the right pane.
Display the memory status Click Memory in the navigation tree to access the memory information and display the status page for memory in the right pane. Related and important counters for memory appear in the Analysis Data table. The explanation details a high average page read rate (reading from disk), and the available physical memory is less than 20 MB. This condition indicates that the system needs more physical memory to handle the load, and this recommendation is made.
This graph displays the last 25 samples taken for the server. The duration of the graph is 25 times the sampling rate of 10 seconds, or 4.16 minutes, as shown near the upper-right corner of the screen. The graph displays what occurred on the server from 1:13 to 1:17 P.M. The pattern of the curves is rising, peaking, and declining, which might be caused by the completion of the new batch job.
Display the controller Click the + icon next to SA 5i in the configuration tree to display the tree structure for the Smart Array 5i controller. The arrays defined for that controller and the SCSI bus entries (ports) appear.
Display the port Click the + icon next to Port 1 Internal Drive Cage in the configuration tree to open the tree structure for the storage system. Scroll the pane to the right to display the physical drives on that SCSI bus. Notice that some drives have a major icon next to them. Display the drive Click the drive ID 0: 146G 15K U320 (Array A) to select. The graph displays the performance statistics for that drive over the last 50 minutes. The graph is trending upward to the right and leveling high.
Display the inventory Select the Inventory tab to display information for the selected drive. Hardware or configuration details for the currently selected device appear. Debugging a network storage bottleneck condition The following sections detail the appropriate actions to take when a bottleneck condition exists on network storage. In this scenario, the server name is “pmpserver,” sharing network storage MSA1500cs.
The problem or problems with a server can also be seen within the configuration navigation pane. The tree structure in the configuration navigation displays the configuration of each node down to the individual components monitored by PMP. The icons used in the tree next to a node or component indicate the performance status for that item or the item under the node. The performance status icon for the selected node is also displayed in the Results frame.
Display the storage status Trace the performance issue by following the amber (Major) icon. Click Storage in the navigation tree to access the storage information and display the status page for storage in the right pane. Related and important counters for storage appear in the Analysis Data table. The Analysis Explanation indicates that at least one storage controller or network storage enclosure is reporting a performance issue.
The Analysis Explanation indicates that the network storage enclosure is experiencing a performance issue. Display the logical drive under network storage Click Disk 4 in the navigation tree to access the network storage information and display the status page for storage in the right pane. Related and important counters for storage appear in the Analysis Data table. The Analysis Explanation indicates that this logical drive is experiencing a performance issue.
The Analysis Explanation indicates that this network enclosure is experiencing a performance issue.
Display the array Click Array B (1 drives) in the navigation tree to access the array information and display the status page for the array in the right pane. Related and important counters for the array appear in the Analysis Data table. The Analysis Explanation indicates that high disk activity is detected. Display the hard disk status Click ID 6: 18G 10K U3 (Array B) in the navigation tree to access the hard disk information and display the status page for hard disk in the right pane.
The Analysis Explanation indicates that high disk queue is detected. This indicates a performance bottleneck on a hard disk in Array B.
Debugging a VM Host and Guest Bottleneck Condition The following sections detail the appropriate actions to take when a bottleneck condition exists on VM Hosts and Guests. In this scenario, the server name is “pmpserver-3,” that hosts the VM Guests. An amber (Major) icon in the PF column indicates that a bottleneck condition exists on the server (VM Host) named “pmpserver-3.” Select the server (VM Host) Click the major icon to display the PMP Online Analysis window for “pmpserver-3.
Display the VM host status Trace the performance issue by following the amber (Major) icon. Click VMWare GSX Server in the navigation tree to access the VM Host information and display the status page for the host in the right pane. Related and important counters for VM Host appear in the Analysis Data table. The Analysis Explanation indicates that the VM Host or at least one VM Guest is reporting a performance issue.
The Analysis Explanation indicates that this VM Host is experiencing a performance issue. Display the VM guest Click, one of the associated VM Guest in the navigation tree to access the VM Guest information and display the status page for the VM Guest in the right pane. Related and important counters for the VM Guest appear in the Analysis Data table. The Analysis Explanation indicates that “At least one component has a critical performance issue.
Display the VM guest processor status Click Processor in the navigation tree to access the VM Guest processor information and display the status page for the VM Guest processor in the right pane. Related and important counters for VM Guest Processor appear in the Analysis Data table. The Analysis Explanation indicates that “Processor utilization is extremely high (over 90%) and probably causing performance degradation.
Performance Issues with no hardware upgrade recommendation The following scenario provides an example of a recommendation that does not suggest additional hardware. In the following figure, a major status icon is displayed for the processors. The recommendation suggests determining if the processing load can be distributed more evenly across the available processors. This recommendation indicates that more than one logical processor exists in the system and at least one processor is not fully utilized.
Performing a static analysis The following sections provide examples of using static analysis. General Usage Performance issues are often introduced during system configuration. This might be caused by a planning error, an error in following the prescribed configuration path, or a result of many updates to a system over time. Regardless of the cause, PMP can analyze static configurations and make suggestions to areas of concern before a change occurs.
Half duplex port Scan through the report shown in the following figure. The first potential issue is indicated by the message “At least one port is running in Half Duplex mode.” PMP indicates that a network port is running half duplex, which can cause a bottleneck. In this case, half duplex was set in error, and it is causing an artificial bottleneck for network traffic over that port.
Unassigned disks The next message that might indicate a problem in the report is “One or more physical disks are not assigned to an array.” In many cases, this condition might be expected, particularly if online spare drives are configured because they are not part of an array set. However, this condition might be unexpected because of an interrupted array configuration attempt.
Reduced SCSI speed The message “At least one drive is running at a reduced SCSI speed on SCSI Port 1” appears in the report. In this example, the HP ProLiant DL380 server was configured correctly with a new Smart Array controller and Ultra-320 drives, but the drives are not running at Ultra-320 speed. Further examination of the hardware reveals an older SCSI cable as a potential problem.
IMPORTANT: When Online Analysis is started for a server, an Online Analysis session is started automatically for any licensed network storage enclosure connected to that server. IMPORTANT: As PMP supports VMware ESX 2.5.2, 2.5.3, 2.5.4, 3.0, and 3.0.1, there are differences in the way PMP analyzes servers running with VMware ESX operating systems.
The following icons are provided for bottleneck navigation: • — Go back to the previous bottleneck • — Go forward to the next bottleneck For example, if a status change happens at 6:00 A.M. from green to amber, yellow to amber, or green to yellow, click the next or previous bottleneck navigation button and the Offline Analysis page for 6:00 A.M. appears. However if status changes at 6:00 A.M.
To generate a report: 1. Click Reports>Performance Management Pack Reports. 2. Click the format in which to generate a report. Available formats are: 3. • Static Analysis Report • System Summary Report • CSV File Generator • Server Availability Report Refer to the following sections for the appropriate report type.
Static analysis report 1. A separate browser window appears listing the selected servers. Select the checkbox next to the server for which to generate a report. 2. Enter the appropriate dates in the Report on Data From and Report on Data To fields to determine the period from which to generate the report. Leave these fields blank to generate a report from the current performance data. 3. Click Generate Report. The report appears in a separate browser window. SQL queries list 1.
System summary report 1. A separate browser window appears, listing the selected servers. Select the checkbox next to the server for which to generate a report. 2. Enter the appropriate dates in the Report on Data From and Report on Data To fields to determine the period from which to generate the report. 3. Enter an appropriate file name with an .htm extension in the File Name field. 4. Click Generate Report. The report appears in a separate browser window. SQL queries list 1.
CSV file generator 1. A separate browser window appears, listing the selected servers. Select the checkbox next to the server for which to generate a report. 2. Enter the appropriate dates in the Extract Data From and Extract Data To fields to determine the period from which to generate the report. 3. Enter an appropriate file name with a .csv extension in the File Name field. 4. Click Generate File. The report appears in a separate browser window. SQL queries list 1.
SQL queries list 1. From the Server Availability Report screen, select the checkbox next to the server for which you need to generate a list. 2. Enter the appropriate dates in the Report on Data From and Report on Data To fields to determine the period from which to generate the list. 3. Click Show SQL queries. The SQL query appears in a separate browser window. NOTE: Server Availability report is not available for the VM Guests.
PMP maintenance This chapter describes routine maintenance tasks involved with PMP. The following topics are discussed: • License Administration Access Levels • Monitoring Administration • Manual Log Purge • Setting Performance Threshold values License administration access levels The Monitoring Administration, Manual Log Purge and Set Performance Threshold pages display information according to the Users and Authorization settings, which are set within Systems Insight Manager.
Licensing Servers To license servers for PMP, use one of the following pages: • Configuration page • Online Analysis page Licensing Servers using the Configuration page To license servers using the Configuration page: 1. From the HP SIM console toolbar, select Options>Performance Management Pack>Configuration. The Verify Target Systems page appears. 2. Select the servers you need to license and click Next.
3. Click Next. The License unlicensed systems page appears. This page displays the number of available licenses and the number of servers that need to be licensed. If the number of servers to be licensed is more than the number of available licenses, you need to add a key using Add Key. For more information on adding a key, see Adding Licenses section later in this chapter. NOTE: When there are no servers available for licensing, the page displays the Previous and Run Now buttons.
Adding Licenses While licensing a server, it is possible that the number of servers that need to be licensed exceed the number of licenses that are available. In such cases, a license or a key must be added to license the additional servers. To add a license: 1. In the HP SIM console toolbar, select Deploy>License Manager. The Product License Information pane appears. 2. Select Performance Management Pack and click Manage Licenses. The Manage Licenses pane appears. 3. Click Add Licenses.
Adding a License from a file Through HP SIM, you can add one or more keys to the database by saving it in a file, and then uploading it into the database. The file must be in the XML format and must be saved with a key extension. The file can contain any number of keys in it. Once the file is uploaded, all the keys saved in it are automatically added to the database. You can create a file and add the keys to it. However, be sure to follow the following format:
Network storage licensing Network storage enclosures must be individually licensed before performance data can be viewed using PMP. In addition, each server attached to the network storage enclosure must also be individually licensed before performance data can be viewed for that server or the server component, including the network storage component.
Licensing cellular Integrity systems The procedure to license cellular Integrity systems varies from the procedure to license other servers. To license the cellular Integrity systems 1. Log into HP SIM. The All Systems page appears. 2. Select the server you need to license and click Option > PMP > Monitoring Administration. The Verify Target System page appears. 3. In the Verify Target Systems page, if you select Next, the Integrity system is automatically licensed.
Changing server monitoring status Servers that are licensed for monitoring are displayed on the Monitoring Administration page with a checkbox next to them. Servers that are currently being monitored by PMP appear with “Started” as the value in the Monitor Status column. Servers that are not currently being monitored have no value in the Monitor Status column. Note the default values for monitoring status and number of samples listed beside each server.
To enable logging for a network storage enclosure: 1. From the Systems Insight Manager console toolbar, select Options>Performance Management Pack Options>Configuration. The Monitoring Administration page appears in a new window. 2. Select the Network Storage tab. Network storage enclosures licensed and currently being monitored appear in the list. 3.
To enable logging for VM Guests: 1. From the Systems Insight Manager console toolbar, select Options>Performance Management Pack Options>Configuration. The Monitoring Administration page appears in a new window. 2. Select the Server tab. Servers or VM Hosts and VM Guests available for monitoring appear in the list. 3. Identify the VM Guest for which to change monitoring parameters from the list, and select or deselect the box next to the VM Guest.
Monitor & Log—Enables monitoring of servers and stores the performance in the PMP database for playback later. • When monitoring is started for a server, the alert and log status in the table will reflect that the Alert and Log options are chosen. When the monitoring is stopped, the table will always show the previous state of Alert and Log status. The following scenario explains the details. Sl # Check box options Monitor Alert Results in the table Log Monitoring Alert Log 1 2 3 4 5 6 7 1.
PMP maintenance 55
Modifying Management Agents from Internet Explorer 1. Browse to the server 2. Log in as an administrator. on which you have installed the Management Agents (for http://servername:2301/, where servername is the name of your server and 2301 is the active port). 3. Select the Settings tab. 4. Click SNMP Configuration>Management Agents. 5. Modify the Data Collection Interval (sample rate) as necessary, and click Apply.
Set Performance Threshold Set Performance Threshold option helps the user in setting performance threshold values for processor, memory, NIC, storage and PCI subsystem of the server(s). You can set the performance by navigating from the following locations: 1. Menu 2. From Monitoring Administration page 3. From Online Analysis page To set the threshold using the Set Performance Threshold menu option: 1.
To set the threshold using the Set Performance Threshold from the Online Analysis Page: 1. From the Online Analysis Page, System Details frame, click the link “Set Threshold …” NOTE: The threshold cannot be set for Network storage and VM Guests. The threshold values are set only for the server being analyzed. 2. Click Apply to set the threshold values. 3. Click Cancel to go back to the online Analysis.
Threshold values The threshold setting for the various subsystems allows adjustment of the analysis of PMP to account for highly variable or stable system performance. A higher value used results in smaller average changes in performance and therefore, generates fewer performance bottleneck indications for a server whose performance changes wildly. A smaller value can be used for servers with more stable performance or when the sampling rate is higher.
Appendix A Sample CSV file PMP enables you to generate many reports, and each user can have a different set of requirements. Therefore, PMP supplies all of the performance information gathered in a file that can be read by many desktop-reporting tools, including Microsoft Excel. The following figure depicts a summary report generated in CSV file format and displayed in Microsoft Excel.
Servers with MSVS and/or VMware GSX (VM Host) • Sample time • Page faults per second • Server performance • Page reads per second • Processor performance • Available megabytes • Average processor busy • NIC performance • Processor busy • NIC megabytes per second • Interrupts per second • Storage performance • Context switches per second • Storage transfers per second • Memory performance • Storage megabytes per second • Hard page faults per second • PCI performance • Pag
Smart Array Controllers • Controller ID • Transfers per second • Name • Megabytes per second • PCI slot • Millisecond per transfer • Sample time • Queue length • Performance Smart Array Logical Drives • Controller ID • Writes per second • Drive array • Read Megabytes per second • Logical drive • Write Megabytes per second • Windows physical disk • Milliseconds per read • Drive ID • Milliseconds per write • Sample time • Queue length • Reads per second Smart Array
Fibre Channel Enclosures • Model • Sample time • Name • Transfers per second • Controller • Megabytes per second • Current role • Milliseconds per transfer • HBA • Enclosure queue • Enclosure SCSI Adapters • PCI Slot • Sample time • Model • Transfers per second • Adapter • Megabytes per second SCSI buses attached to SCSI Adapters • Model • Performance • Adapter • Transfers per second • Bus • Megabytes per second • Sample time • SCSI bus utilization percent S
Host Buses • Device • Performance • Name • Megabytes per second • Sample time • Bus utilization percent Appendix A 64
Appendix B Sample system summary report System Summary reports are created in a browser window in a printable format. The report has two sections. The first section consists of a table showing what percentage of time each server subsystem was in a bottleneck state during the selected interval. This table provides a convenient method for gauging the performance health of a server. The second section of the summary report includes a detailed system configuration listing.
Bus: 1,ID: 0 9.1-GB Pluggable Wide Ultra2 SCSI 10,000-rpm Hard Drive (1") (Part of Array: A) Bus: 1,ID: 1 9.
Secondary Bus Compaq NC3131 Fast Ethernet NIC (Slot # : 5 ) (Empty) (Slot # : 6 ) Compaq Storage Works Fibre Channel Host Bus Adapter/P (Slot # : 7 ) (Empty) (Slot # : 8 ) Compaq Smart Array 4200 Controller (Slot # : 9 ) Tertiary Bus Compaq Smart Array 4250ES Controller (Slot # : 10 ) (Empty) (Slot # : 11 ) System summary report for a VM Guest Performance Management Pack Summary Report for 131.111.1.
Overall performance The overall performance status for a system can be determined by referencing the following values shown in the first half of the summary report: • %Normal—Percentage of overall performance that is satisfactory with no impending or existing bottleneck condition • %Minor—Percentage of overall performance that might be approaching a bottleneck condition • %Major—Percentage of overall performance that is in a bottleneck condition The following table displays configuration details and p
Appendix C Sample Static Analysis Report A Static Analysis Report is an analysis of the hardware configuration for the server as a whole, identifying potential problem areas. A complete list of the configuration information analyzed to report the status of server subsystems is provided in Appendix A of this guide.
Appendix D Sample server availability report Server Availability report is created in a browser window in a printable format. The report has one section, consisting of a table showing what percentage of time each server was in a bottleneck state during the selected interval, Accessible Duration of Server in hours, Inaccessible Duration of Server in hours and Average of the above Metrics for selected servers. This table provides a convenient method for gauging the availability of a server).
Index A HP Systems Insight Manager. See Systems Insight Manager access level, 44 administration: access level, 44; monitoring, 50 array, displaying status, 26, 30 Auto Refresh, 13 I B L bottleneck: debugging, 16, 21, 28; navigation, 38 license: demo, 50; free, 50; network storage, 49 logged files, deleting, 14, 57, 59 C comma-separated-value file.
