Overview of Internetwork Performance Monitor

This chapter provides an overview of the CiscoWorks Blue Internetwork Performance Monitor (IPM) application. It contains the following sections:

What Is IPM?

IPM is a network management application that allows you to monitor the performance of SNA and IP traffic in a routed network. IPM is designed to measure and analyze response times on a hop-by-hop (router-to-router) basis within the network.

You can use IPM to do the following:

Troubleshoot problems by checking the time delays between devices
Send SNMP traps and Systems Network Architecture (SNA) alerts and resolutions when a user-configured threshold is exceeded, a connection is lost and reestablished, or a timeout occurs
Analyze potential problems before they occur by accumulating statistics, which can then be used to model and predict future network topologies

Note The IPM system is composed of two parts, IPM network management application and the RTR feature of Cisco IOS software. The focus of this document is the IPM network management application. In some cases, however, it is not possible to fully describe IPM without including information about the Cisco IOS RTR feature. Therefore, we have included some information about the Cisco IOS feature. If there is a discrepancy between the information in this document and that in the Cisco IOS documentation, the Cisco IOS documentation should take precedence.

How Does IPM Work?

IPM measures response times between a source router and a target device. The target can either be an IP-addressable device (a router or workstation) or an IBM Multiple Virtual Storage (MVS) mainframe (SNA response time only, running an IPM VTAM application called NSPECHO). There are two types of measurements that you can take: Echo and PathEcho.

Echo measures the total response time from the source router to the target device.
PathEcho measures the total response time as well as the incremental response time for each hop in the path between the source router and the target device. PathEcho is only for the IP protocol.

The IPM application is used to configure the RTR agent in each source router and then extract and display the response-time information. The RTR agent in the router takes the actual response-time samples between itself and the target device. The IPM application normally extracts the response-time data every hour from each source router. There is also a real time feature that allows you to immediately display the response-time data.

Measuring Response Time in IP Networks

In an IP network, you can request either an Echo or a PathEcho measurement.

If you request an Echo measurement, the Cisco IOS RTR feature in the source router issues an Internet Control Message Protocol (ICMP) ping to the target device and extracts the response time data from the response.

If you request a PathEcho measurement, the Cisco IOS RTR feature first issues a traceroute to determine the path through the network from the specified source router to the specified target device. The data returned from the traceroute contains the host name or IP address of each of the routers in the path. RTR then issues ICMP pings to each of the routers listed in the traceroute data. The ICMP ping returns statistics regarding the response time between the specified source and each of the routers.

Note An ICMP ping has a low priority, which ensures that the worst possible response time is being measured.

Measuring Response Time to an SNA Mainframe

For SNA environments, IPM allows you to measure the response time to an MVS mainframe using the following types of SNA sessions:

SSCP-LU
LU0

Because SNA is a connection-oriented protocol, the only type of measurement you can request is Echo.

When measuring response time to an SNA mainframe, IPM uses an "SNA ping." The source router sends a block of data (a request) to an IPM-supplied mainframe echo program, NSPECHO. NSPECHO responds with a block of data (a response), which is used to determine the response time.

You can customize both the request and response sizes so that traffic flow for various applications can be modeled.

Defining a Collector

To measure and display response-time data, use IPM to define a collector agent in the source router. A collector is a definition of the source router, the target device, an operation (the type of measurement desired such as ECHO, IPECHO, SSCP-LU, or LU0), and the frequency and time that the collector is to run.

Components of IPM

The IPM system is modular in its design (Figure 1-1). Some of the components are provided as part of the IPM application, others are provided as part of the Cisco IOS Release 11.2 software. The RTR feature of Cisco IOS Release 11.2, which is available only in the IBM or Enterprise feature set, contains the Cisco RTTMON MIB and the SNMP agent.

The components of IPM network management are described in the following sections:

Figure 1-1: Components of IPM

IPM Configuration Process

The IPM configuration process runs on the UNIX workstation and is used to configure and schedule IPM collectors on source routers. It is provided as part of the IPM network management application. The main functions of the configuration process are as follows:

Configuring IPM source routers
Configuring IPM targets, SNA hosts or other IP devices, in the IPM database
Configuring and managing the IPM collectors in the network routers
Sending the collector configuration data to the source router
Notifying the IPM data collector to start collecting data for new collectors
Starting the IPM display process so you can view the response-time results sent back from the collectors
Dynamically enabling or disabling logging of the servers

SNMP Server

The SNMP server resides on the UNIX workstation and is provided as part of the IPM application. The IPM configuration and IPM data collection processes use SNMP to deploy collectors in the routers and to gather data from the RTTMON MIB.

IPM Collector

An IPM collector is a user-defined entity on the source router that includes information about the target device, the protocol to be used for measuring response time, the frequency at which the response time is measured, and the amount of time for which response-time measurements should be taken.

The purpose of an IPM collector is to capture statistics and error information from the IPM-enabled routers. By default, the following information is captured and collected:

Minimum and maximum response times
Number of completions
Sum of completion times
Sum of the squares of completion times
Accumulation of errors for noncompletions
Total attempts (errors plus number of completions)
Statistical distributions of response times

Note Do not use the Cisco IOS RTR commands to access and change IPM collectors that were created by the IPM application. Changes made using Cisco IOS commands might render the changed collectors unusable by the IPM application.

SNMP Agent

The SNMP agent resides in the source router and is provided as part of Cisco IOS. The SNMP agent receives requests from the SNMP server to deploy the IPM collectors.

RTTMON MIB

The RTTMON MIB is a proprietary MIB created by Cisco to obtain and store round-trip time statistics. The IPM application obtains the round-trip time statistics from this MIB. For additional information about this MIB, see the "IPM MIBs" appendix.

IPM Data Collector

The IPM data collection server resides on the UNIX workstation. It is provided as part of IPM. The data collector process performs the following:

Reads all collector configurations from the Open Database Connectivity (ODBC) compliant database. This is done at startup and whenever the configuration process indicates that a new collector is added.
Collects response-time data for each collector from the network routers. The data collector process uses the SNMP server to collect the data.
Stores the response-time data in the ODBC database for each collector.
Informs the display process that there is new response-time data to display.

Relational Database

If you are using the Sybase database that comes with CiscoWorks, the relational database can reside on either the network management workstation or on an accessible database server. Otherwise, the relational database must reside on the network management workstation. Several of the IPM processes interact with the relational database, for example:

The configuration process stores all source and collector configurations in this database.
The data collector stores the response-time statistics in this database.
The display process displays IPM statistics from this database.

IPM Display Process

The IPM display process runs on the UNIX workstation to display the response-time statistical information associated with an IPM collector. The display process, started by the configuration process, reads the information from the database. When the data collector process indicates a new collection, the display process informs you to update the display. If Auto Refresh is enabled, then the display is automatically refreshed.

The IPM display process includes the capability to do the following:

Start a "Go Live" session between a source and a hop (or target)
Start an end-to-end session between a source and target
Refresh the paths
Automatically refresh the paths
Demand poll
Print the displayed charts and graphs

Mainframe Echo Program

When measuring response time to an SNA mainframe, IPM measures round-trip response times between a source router and an echo program in an SNA mainframe. A program in the SNA mainframe is responsible for providing the echo back to the router. IPM provides a mainframe echo program, NSPECHO, that you can install on an MVS mainframe for this purpose.

How Often Statistics Are Gathered

Each collector's statistics are retrieved once each hour from the source router. However, each collector may measure the response time between the source router and the target device more often. When you use the IPM configuration process, you specify the interval at which each measurement operation is performed by a collector.

Table of Contents