The Java Performance Manager (JPM) component of Nways Manager Element Manager allows you to store, display, and analyze the values of counter and gauge MIB variables on devices being managed by a Java-based Management Application. The JMA can be specifically created for a particular device, or it can be the generic JMA that can manage standard MIBs on any SNMP agent.
| Overpolling |
|---|
|
Polling a large number of MIB objects from a device in a relatively short period of time, or polling a large number of MIB objects for a sustained time may result in overpolling of the device and cause agent degradation. If a device agent's operation appears degraded, then reduce the number of objects being polled, increase the polling interval, or stop polling for the device. This will indicate if the condition is a result of overpolling. |
The Java Performance Manager is a set of functions designed to provide performance management for any Java-based Management Application. The JPM is not a stand-alone product, but rather a set of functions that are integrated with the JMA.
The JPM uses Distributed Intelligent Agents (DIA) and a Distributed Performance Manager (DPM). Any number of DIAs can be distributed in an IP network to provide performance monitoring on local devices. A DIA performs the following tasks:
A DIA will reestablish its connection to its DPM after a program or network outage, allowing the DPM to restore the DIA's prior program state.
You can specify the variables to monitor, and create and customize performance reports that can be viewed with any Web browser.
The Distributed Intelligent Agent is a Java program that performs SNMP polling for the Java Performance Manager. A DIA can run on any platform that supports version 1.1.6 of the Java run-time environment.
One registered DIA is included with Element Manager. For information on purchasing additional Registered Agent Accesses, which are required for additional DIAs, please see page ***.
To install a DIA on a workstation other than the Nways Manager workstation, see the readme.txt (/usr/CML/JMA/dia for AIX and HP-UX) (\Nways\CML\JMA\dia for Windows NT). A DIA is automatically installed on the Nways Manager workstation. If you do not configure any other DIAs, the instance of the DIA on the Nways Manager workstation will poll all nodes managed by JMAs in Nways Manager. All DIAs are configured through the DpAdmin tool on the Nways Manager workstation. See the section, "JPM Server Configuration", for more information on configuring DIAs.
In addition to polling, DIAs can perform thresholding operations on polled variables. You can explicitly configure JPM thresholds, or you can elect to have them set by JPM based on the mean and standard deviations of previously polled values. By default, the threshold for all objects polled by JPM is 3 standard deviations above the mean, calculated after 20 polling intervals and continuously updated as the mean and standard deviation change.
When the DIA detects that a configured threshold has been exceeded, it notifies the JPM server. If a JMA is running for the agent that caused the threshold to be exceeded, then a Performance bullet on the navigation tree of that JMA will change color to red. When the value returns below the threshold, the Performance bullet will return to green. These performance status changes are propagated up the navigation tree, causing higher branches in a tree to be red, yellow, or green, depending on the composite status. For more information on the connection between the Performance bullet on the navigation tree and JPM polling, see "JPM Server Configuration".
JPM uses a Java Database Connectivity (JDBC)-compliant database to store the historical data that it captures.
IBM has successfully tested the JPM with DB/2 Universal Database.
To install and configure DB2 for use with JPM:
and at least 100 MB of free disk space.
to act as the owner of the Administration Server.
mount /dev/device_name /cdrom
where device_name is the name of your CD-ROM drive.
cd /cdrom
db2setup
chown nwaysdb2:db2asgrp /home/nwaysdb2/.profile
. sqllib/db2profile
chown nwaysadm:db2asgrp /home/nwaysadm/.profile
. sqllib/db2profile
export DB2DIR=/opt/IBMdb2/V5.0 export DB2INSTANCE=nwaysdb2 export SHLIB_PATH=/home/nwaysdb2/sqllib/lib export PATH=$PATH:/home/nwaysdb2/sqllib/bin:/home/nwaysdb2/sqllib/adm:/home/nwaysdb2/sqllib/misc
db2set -i nwaysdb2 DB2AUTOSTART=YES
rcdb2:2:once:/etc/rc.db2 > /dev/console 2>&1 # Autostart DB2 Services
The first field ("rcdb2") in the line can only be 4 characters long. The second field ("2") is the run level in which this entry is to be processed. A blank run level field means that the entry is valid for all run levels. To ensure that DB2 is started on reboot, change the line to the following:
rcdb::once:/etc/rc.db2 > /dev/console 2>&1 # Autostart DB2 Services
db2start
db2 CREATE DATABASE IBMNMPDB
You can find information about unzip at:
http://www.cdrom.com/pub/infozip/UnZip.html.
/usr/CML/JMA/bin/dpadmin
| Note: | When you run Performance Management Configuration, a graphical window is displayed. To use this window, you must have X-access to your console. |
When first installed, JPM does not use a database. A limited amount of historical data is stored in memory, and this data is lost when new data is received. To start storing information in the database, you must specify the database location and drivers using the DpAdmin tool. For information on how to do this, see JPM Server Configuration.
You must ensure that you have adequate free space on the database file system before starting database storage with JPM. The amount of space needed is highly dependent on the number of agents being polled and the polling interval. As a starting point, polling 20 agents with the default 20 minute polling interval, a 250 MB file system could be filled up in three to four weeks of continuous polling. It is recommended that you monitor the amount of storage consumed by the JPM database for several weeks after starting use of the database, and develop a database maintenance plan based on your observations.
To purge old data from your database, run the JPM Database Maintenence application by selecting IBM Nways Java -> Database Maintenance from the NetView Tools menu, or by issuing the following command:
/usr/CML/JMA/bin/dbmaint.sh
Using the JPM Database Maintenance program, you can specify a time period for which you want to retain data, and then choose to:
or
| Important |
|---|
|
If you chose not to have DB2 started automatically when the machine is started, then you will have to issue the db2start command from the MS-DOS command prompt before creating a database. |
After making sure that the JDBC drivers are installed and DB2 is started, follow these steps to create a database:
By default, a database will be created on the same disk that DB2 was installed. It is advised that this disk partition have at least 60 to 100 MB of free disk space. If you intend to store historical statistics for more than 20 devices over a period several weeks or longer, then you will need approximately 1 GB or more of available disk space.
If you are using HP OpenView Network Node Manager on Windows NT Version 5.01, then you may be asked to enter your username and password. If so, enter the administrative password that you created at the beginning of this process.
If you encounter problems connecting to the JDBC-enabled database from
Nways Manager, try the following solutions.
| Message: | Could not find JDBC driver classes. Closing connection. |
| Action: |
Install or reinstall the JDBC drivers.
|
| Message: | JDBC connect error. Check database URL and make sure your username and password are correct. Closing connection. |
| Action: |
|
After a period of use, it may be necessary to clear out the DB2 database. To clean out the database, change to the directory nways\bin, where nways is the directory into which Nways Manager was installed, and run the command: db2cmd CreateDatabase.bat -create
You can customize the MIB objects that are presented to the JMA when a Performance bullet is selected from the JMA navigation tree. Element Manager makes default selections that provide basic information about the interfaces, protocols, and system elements in the navigation tree. Some MIB counters and gauges that might be of use in some network management situations are not polled as part of the default configuration. You can use the DpAdmin tool to customize what data is displayed under a Performance bullet in the JMA navigation tree.
This section explains the JPM data concepts and offers tips for using the DpAdmin tool to modify the default JPM customization.
JPM organizes MIB objects into these groups:
The default configurations of these objects are stored in text files found in the directory install_dir/java/properties/startup.
Important
It is often useful to customize two aspects of JPM performance that affect all these default objects by modifying these text files prior to beginning JPM polling.
The most commonly modified properties are:
To modify the polling interval, use the following procedure:
Later, you can change the polling interval for individual polling objects using the DpAdmin tool. Remember that the total load placed on DIAs by polling and the volume of SNMP traffic generated in your network is increased by a smaller polling interval.
To modify the view length, use the following procedure:
To load a new default configuration, follow these steps:
/opt/OV/bin/ovstop JMAintegrator
/opt/OV/bin/ovstart JMAintegrator
After you have loaded the default JPM configuration, you can use the DpAdmin tool to modify the JPM configuration. You can launch the DpAdmin tool in two ways:
The tabs on the DpAdmin notebook allow you to configure a wide variety of behavior associated with JPM. The tabs are:
With previous releases of Element Manager, JPM automatically started polling a device when a JMA was launched for that device and continued polling until you instructed JPM to stop polling. In Element Manager Version 2.1, JPM temporarily polls devices while a JMA is open, but permanently polls only those devices and device types that you specify on this page.
The Performance Analyzer application can be used to retrieve performance data from the JPM database, construct graphs, and create reports. There are three ways to start the Performance Analyzer application:
The Performance Analyzer application gets all of its statistical data from the Nways Manager Performance Database. By default, this is IBMNMPDB. To use the Performance Analyzer application, you must configure Nways Manager to store its performance data to a database.
The Managed Devices branch of the navigation tree contains an icon for each device in the Nways Manager Performance Database. If the Performance Analyzer was launched from a Java-based Management Application, you will see only that device under the Managed Devices tree.
Under each managed device, you will see a magnifying glass icon representing a performance object. A performance object is an SNMP MIB variable or expression that Nways Manager is polling for that device.
Under each performance object, you will see each instance that was discovered for this device. Single-instance (scalar) performance objects will only have a single instance ("0") while multi-instance (tabular) performance objects can have several.
The Analysis Reports branch of the navigation tree contains an icon for each user-created report.
To create a graph, you must initialize the graph, set the time range, add elements, and modify the attributes of the graph or graph elements.
The Clear button is used to clear all elements from the graph and to reset the graph times to the range indicated by the Time Range control. Notice that the graph will not accept new Time Range values until it is cleared.
The Time Range control gives you the ability to manage the amount of time (from the current time into the past) that you want your graph to cover. Graphs with larger-than-needed time frames can be slow to retrieve data from the database and the graph can become cluttered, so be sure to choose the smallest time range that meets your needs.
Now that your graph is initialized, you can begin to add graph elements to it. A graph element can be added by double-clicking on a performance object instance or by selecting one or more instances and clicking the Edit button. The selected instances will be added to the graph with the instance details displayed in the legend below the graph.
You can modify the attributes of a graph or of a graph element. You can modify graph attributes by double-clicking on the top row of the legend below the graph. You can modify graph element attributes by double-clicking on the desired element in the legend.
You can manipulate a graph by:
You can toggle between a pie and bar chart view of the data. The same zooming, translating, and resetting options are available, but now an explode operation will display the value of the selected element.
After creating a graph, you can save your changes to a report. Clicking Save will bring up the Save Analysis dialog. Enter a unique name (up to 18 characters with no spaces) and click Yes to store the graph in the performance database as a report. A new report icon will be created under the Analysis Reports branch of the navigation tree.
You can launch a report that you have created by selecting a Time Range and then double-clicking on a report icon (or selecting one or more reports and clicking the Launch button). If you want to make changes to a report, simply select the report and click Edit to load the report as a graph. Make your changes to the graph and click Save to store the graph as a report.
The Real-Time Poller/Grapher application allows you to view a real-time graph of performance data. The data is polled more often than ordinary JPM polling, but the collected data is not stored in a database. Data for multiple MIB objects and from multiple devices may be combined on a single graph.
To start the Real-Time Poller/Grapher application, select Tools -> IBM Nways -> Real-Time Poller/Grapher from the OpenView menu.
When it is first launched, the Real-Time Poller/Grapher asks for the IP address of a device to poll, a community name, and initial polling interval. The Poller/Grapher navigation tree displays all Performance Objects that are available, even though some of them may not be supported by the initial device. Expanding the navigation tree for a Performance Object shows the instances of that object on the device.
To begin polling:
A graph appears in the upper-right portion of the window. The graph is updated each time a new data point is received.
| Note: | It may take two polling intervals for the first point to appear on the graph, since the value for many Performance Objects can only be calculated by comparing two consecutive data points. |
You can add another device to the navigation tree by pressing the Add Host button. All Performance Objects currently being polled will continue to be polled. This allows you to compare values of two or more devices on the same graph.
To stop polling any Performance Object, select it in the list of polled objects and press the Stop button.
If the Performance Analyzer application detects that a Managed Device contains Performance Objects related to TN3270E response time monitoring, specific grouping for TN3270E will be created in the navigation tree. This section describes the specific functions available for TN3270E response time monitoring.
TN3270E groups are represented by a mainframe/terminal icon. Under each TN3270E group are the clients that belong to the group. A client is identified by an IpAddress:Port label. If the response time monitoring is aggregated for all clients in the group, a single aggregate client will appear under the group.
Selecting a TN3270E client and clicking the Edit button will load the TN3270E Response Time Monitor graph. This graph contains five elements corresponding to the five response time "buckets" defined for this TN3270E client. The buckets separate all TN3270E transactions into response time ranges. The TN3270E Response Time Monitor graph shows the number of transactions that fell into each range.
Once the graph is loaded, you can make changes and save your changes just like any other graph.
Double-clicking on a TN3270E client (or selecting one or more clients and hitting the Launch button) will launch the TN3270E Response Time Monitor report.
If the Performance Analyzer application detects that a Managed Device contains Performance Objects related to routers (CPU, memory, and buffers), an entry for Routing will be created in the navigation tree. This section describes the specific functions available for managing routers.
Selecting a Routing icon and clicking Edit will load the Router Utilization graph. This graph contains the CPU, Memory, and Buffer Utilization for this router.
Once the graph is loaded, you can make changes and save your changes just like any other graph.
Double-clicking on a Routing icon (or selecting the icon and clicking the Launch button) will launch the Router Utilization report.
If the Performance Analyzer application detects that a Managed Device contains Performance Objects related to frame-relay circuits (utilization, traffic, and congestion), entries for each frame-relay interface will be created in the navigation tree. This section describes the specific functions available for managing frame-relay circuits.
Frame relay interfaces are represented by a wire/connector icon. Under each frame relay interface are the circuits that are defined to that interface. A circuit is identified with its DLCI number as its label For each circuit, icons for Traffic, Utilization, and Congestion are defined.
Selecting either Traffic, Utilization, or Congestion and clicking Edit will load the appropriate frame-relay graph.
Once the graph is loaded, you can make changes and save your changes just like any other graph.
Double-clicking either Traffic, Utilization, or Congestion (or selecting any combination and clicking the Launch button) will launch the Frame Relay Circuit report.
Java Performance Manager's reporting capability provides access to the historical data that is stored in the JPM database without having to navigate through the JMA. It provides this access through Java applets that are imbedded in HTML pages made available through a server on the Nways Manager workstation. There is one type of report that provides graphic displays of stored historical data, and two types that provide text-based displays.
To simplify Web access to the reports, a Report Catalog HTML page that provides links to all reports is automatically created and updated by JPM. The page name is:
install_dir/java/websvr/reports/ReportCatalog.html
where install_dir is the directory where Element Manager is installed.
The simplest way to create a report is to click on Add View to Report button when viewing a graph in a JMA. When the report creation dialog is displayed, type in the name for the report you want to create (JPM automatically adds .html to create a filename), and select a timespan for the report. You can create a report that covers the number of hours, days, weeks, or months that you specify.
To create a simple report that displays the information in a graph, use the default values for the Report Type field and Save as Applet checkbox, and leave the Prompt for Additional Hostnames box unchecked.
Performing these steps will create an HTML file with APPLET tags and parameters. The JPM Chart applet communicates with the JPM server to retrieve the historical data identified by the parameters and displays the data in a chart. A report created with default parameters through the Add View To Report button will display information only from the agent that the JMA is open for, and for all instances of tabular MIB variables. Report Parameters describes how to create reports that have different parameters.
You can create reports that display selected rows of tabular MIB data (such as interfaces), or that combine historical data from multiple agents on a single graph. This section describes how to code parameters in HTML files to create these variations of reports.
To modify the reports created by JPM, use any text editor, or select Edit Reports from the JPM graph displayed when you select a Performance bullet from the JMA navigation tree.
To create a report that has a single instance of a tabular MIB variable for a single agent, specify a value for the instance parameter that is created by default in the report file. The value for an instance parameter has the format:
MIB_variable_name=value
For example, to create a report that shows the Interface Utilization for interface number three, first use JPM to create a report that shows Interface Utilization for all interfaces, then modify the parameters as shown:
<param NAME=number_of_nodes VALUE=1> <param NAME=host_name_1 VALUE="10.10.3.100"> <param NAME=view_name VALUE="Interface Utilization"> <param NAME=instance VALUE="ifIndex = 3">
Some tables require two MIB variables to identify a particular row in the table. For example, to create a report that shows the Frame Relay Traffic for circuit 17 on interface 4, modify the instance parameter as shown:
<param NAME=number_of_nodes VALUE=1> <param NAME=host_name_1 VALUE="10.10.3.100"> <param NAME=view_name VALUE="Frame Relay Traffic"> <param NAME=instance VALUE="ifIndex = 4, frCircuit = 17">
To create a report that shows more than one instance (but not all instances) of a tabular MIB variable for a single agent, create a host_name and instance_parameter pair for each instance you want to view. In this case, you must set the number_of_nodes parameter to the number of instances you want to see on the graph:
<param NAME=number_of_nodes VALUE=2> <param NAME=host_name_1 VALUE="10.10.3.100"> <param NAME=view_name VALUE="Interface Utilization"> <param NAME=instance_1 VALUE="ifIndex = 4"> <param NAME=host_name_2 VALUE="10.10.3.100"> <param NAME=instance_2 VALUE="ifIndex = 5">
To create a report that shows instance of MIB variables from more than one agent, create a host_name and instance_parameter pair for each agent you want to see. Also, change the number_of_nodes parameter to the number of host/instance pairs you create. For example:
<param NAME=number_of_nodes VALUE=3> <param NAME=host_name_1 VALUE="10.10.3.101"> <param NAME=view_name VALUE="Interface Utilization"> <param NAME=instance_1 VALUE="ifIndex = 4"> <param NAME=host_name_2 VALUE="10.10.3.102"> <param NAME=instance_2 VALUE="ifIndex = 5"> <param NAME=host_name_3 VALUE="10.10.4.101"> <param NAME=instance_3 VALUE="ifIndex = 9">
JPM will create reports that contain an ARCHIVE parameter on the APPLET tag. This parameter tells your Web browser to load Java class files from the archive specified, in this case ReportClasses.jar.
To improve performance when viewing reports remotely using a Web browser, leave the ReportClasses.jar file in the install_dir/java/websvr/code directory, where install_dir is the directory where Element Manager is installed, and allow the browser to download the file across the network.
There are two types of text-based reports that can be created by selecting Add View to Report from the JMA chart screen, a Single Node Analysis report, and a Multiple Node Analysis Report.
The Single Node Analysis report presents a statistical analysis of all performance objects associated with the selected View. The mean, standard deviation, and recorded high and low values are displayed in a table.
The Multiple Node Analysis report presents a statistical analysis of a single performance object across one or more agents. The mean, standard deviation, and recorded high and low values are displayed in a table.
To create a text report, select Add View to Report while viewing a performance graph in a JMA that contains the performance objects you want to analyze. Choose a name for the report, and change the Report Type field to either Single Node Analysis or Multiple Node Analysis.
You can view reports with a Web browser that supports version 1.1 of the Java Development Kit (JDK). Some popular browsers and the levels that support JDK 1.1 are:
http://java.sun.com/products/hotjava/
When viewing a report, the time period that the report displays is governed by the parameters you selected when the report was created, shown in the time_range and time_length parameters in the HTML file. The starting period for the time range is dependent on the units specified by the time_range parameter. For a unit of HOURS, the time range starts at the beginning of an hour. For a unit of DAYS, the time range starts at the beginning of a day (12:00 midnight). For a unit of WEEKS, the time range starts at the beginning of a week (Sunday by default, but you can change this under the Report Options tab of the DpAdmin tool).
When viewing a report, you can change the time span the report covers by selecting New Time Range. You can specify a range as a certain number of hours, days, weeks, or months, or you can give an exact starting and ending time.
To access the Java Performance Manager, expand the Java-based Management application navigation tree until you reach the balls labeled Performance in the folders. Click on a Performance ball to see the performance information for that item. For more information on using the navigation tree, see Java-based Management Application Window Operation.
Use the JPM to perform the following actions for the resource associated with the Performance ball:
The JPM window gives performance information for the selected resource. It is presented in a notebook-style format.
The tabs at the top of the window correspond to views configured for particular resources. For example, if the resource is an interface, the default views would be Interface Utilization, Interface Packet Quality, and Interface Packet Breakdown. Views present information about the selected resource in graphs created according to preselected criteria.
This section contains some examples of how to perform common tasks using JPM.
Using JPM to see IP performance information for a Cisco router
Configuration
Communication
Protocols
TCP/IP
IP
Monitoring memory and CPU utilization for an IBM 2216 Nways Multiaccess Connector
To monitor memory utilization:
Configuration
System
Environment
Memory
To monitor processor utilization:
Configuration
System
Device
Processor
Creating a custom report using JPM
This example shows the steps to create a custom report showing CPU and memory utilization for several IBM 2216 Multiaccess Connectors.
<param NAME=instance...>
to:
<param NAME=instance_1...>
<param NAME=host_name_n VALUE="z.z.z.z"> <param NAME=instance_n VALUE="">
Replacing n in the example with the ascending integer values starting with 2 (using the same value for both parameters), and replacing z.z.z.z with the IP address of the 2216 you want to monitor. Do this for each set of applet tags.
<param NAME=number_of_nodes...>
so that the value is the same as the number of host-name/instance pairs you have created.
When creating reports for variables that are part of a table, such as variables in the MIB2 interface table, you can see all entries in the table by leaving the instance_parameter variables without values. To see only specific entries in the table, you must specify a value for the instance_parameter by giving the index variable name and the desired value. For example:
<param NAME=instance_1 VALUE="ifIndex=5">
Incorporating a report into a Lotus(R) Freelance Graphics(R) presentation