Configure Transcend to monitor your network most effectively, identify when thresholds are exceeded, and alert you to problems or potential problems.

Monitoring Devices

For Transcend to monitor your devices:

• Use your platform's autodiscovery feature to detect all manageable devices on your network and to create a network map. Transcend NCS applications use this data for their operation. For Transcend NCS applications to recognize 3Com devices from the platform, the device icons must be 3Com device icons.

• Add 3Com devices to an inventory database using Transcend Central. You can import devices from your platform's database. The Transcend Central database defines the devices that many of the Transcend NCS applications manage and allows you to group devices for easier management and faster troubleshooting.

• Create logical and physical groups of the devices in your database using Transcend Central.

Thresholds are the upper and lower limits that you set for the network conditions and events that you are monitoring with network management software. When these limits are exceeded, the management software reports that a threshold has been exceeded (usually by icons changing color). Alarms add to this reporting functionality by allowing you to configure an action to be taken (such as disabling ports or sending e-mail) if the threshold is exceeded.

Alarms that are configured correctly can prevent inconvenient or even catastrophic network failures. The main advantage of alarms is that you can specify at exactly which point an action should take place, and you can tailor them to suit the normal operating conditions of your network.

The first time that you use the Transcend NCS applications, use the default thresholds to see how they apply to your network. After you assess your network's normal behavior, you can adjust the thresholds and alarms to make them more useful for your particular network. See "Identifying Your Network's Normal Behavior" for more information.

Setting Thresholds in Status Watch

You can set a rising threshold and a falling threshold for most Status Watch tools. The rising threshold triggers a status severity change when the threshold is exceeded. The falling threshold causes a status severity change when the excessive activity or abnormal condition has returned to normal.

For example, your Ethernet network may normally accommodate 50 percent utilization. If it exceeds 60 percent for an extended time, your network slows considerably. You want to know when and for how long your network exceeds the threshold of 60 percent.

Status Watch also allows you to set status severity levels for events in the FDDI Status and the System Status tools. You can set the severity level setting for the conditions and events. For some conditions and events, you can specify severity level settings for the individual values of the variables.

For more information about setting thresholds in Status Watch, see the Status Watch User Guide and Status Watch Help.

Setting Thresholds and Alarms in LANsentry Manager

Much of network management involves monitoring for specific network events. With LANsentry Manager, you can specify these events in advance and then know as soon as they occur. This process is known as setting alarms.

Consider the following examples of alarms:

• Example A: The router on your network, which is capable of forwarding data at 3,000 packets per second (pps), appears to have problems forwarding at the top of its specification. You configure an alarm to notify you as soon as the traffic approaches this rate.

• Example B: Your network is running at 1,400 pps. Typically, a Cyclic Redundancy Check (CRC) rate of more than 1 percent of network traffic is considered excessive. You configure an alarm to notify you as soon as the CRC rate climbs above the threshold of 14 pps.

Over time, you build up a library of alarms for your own network.

Refining Alarm Settings

You can refine your alarms for more exact monitoring by setting the hysteresis zone and defining Start and Stop events.

Hysteresis zone

For more control over the conditions that trigger an alarm, you can also specify a hysteresis zone around the specified value. The hysteresis zone ensures that alarms are not triggered due to small fluctuations around the threshold value. The hysteresis zone is the area where a value has fallen below the upper threshold (also called the rising threshold) but has not yet reached a lower threshold (also called the falling threshold). After a rising threshold generates an alarm, the value must fall below the falling threshold before another alarm is generated. For alarms that are set on falling thresholds, the rule is reversed. Figure 7 shows an example of this alarm mechanism.

Figure 7 Alarm Triggering Mechanism

Stop and Start events

In addition to using alarms on their own, in LANsentry Manager, you can use them as Start or Stop events when capturing packets with the Capture application. In Example A, you can start capturing all packets the router transmits whenever the traffic rate rises above 2,800 packets per second and then stop capturing when it drops below this level. In this way, you can capture packets leading up to the event and immediately after. By combining alarms and the Capture application, you have powerful troubleshooting capabilities.

For more information about setting alarms with LANsentry Manager, see the LANsentry Manager User Guide and Help.

Setting Alarms Based on a Baseline

When you determine the baselines of your network's normal activity with Traffix Manager, you can use the Alarms View in LANsentry Manager to set alarms that trigger when network activity deviates from the baseline. See "Baselining Your Network" for more information.

When determining the baseline for setting utilization alarms, use either of these approaches:

• Set alarms for any peaks in network utilization - Pick a baseline value that covers most of your network traffic, ignoring any obvious one-time-only peaks. For example, as users log on at the start of the day, you see a large peak in network utilization. The alarm is triggered whenever such peaks occur.

• Set alarms for exceptional peaks in network utilization - Pick a baseline value that covers the highest possible peak seen when service was still provided. The alarm is triggered at levels higher than this peak, alerting you to the most serious utilization on your network.

When you choose the baseline for error alarms, pick the lowest possible baseline so that the alarm is triggered by any peaks.

Other Tips for Setting Thresholds and Alarms

For SNMP traps to be effective, their thresholds must be high enough so that they do not generate false alarms. On the other hand, high thresholds also mean that small amounts of errors can escape detection. A very small error rate that regularly occurs (such as four per minute) can cause major problems with protocols with large retry delays. For example, some MAC-level errors corrupt packets so that a switch does not forward them.