Monitoring performance and alarms

Using the Management Station performance and fault managing features, you can quickly assess if your system is performing optimally: for example, if performance thresholds have been breached, or if any services are behaving abnormally.

This topic describes how to:

• Monitor system performance: View the operational status of managed hosts and services.
• Monitor vital signs: View performance statistics on various aspects of the system every several seconds.

For information on viewing shared-memory data on hosts configured to run multiple recognition services, see Accessing shared memory details.

Monitoring system performance

Monitoring system performance consists of two basic activities: measuring performance, then analyzing it.

Measuring performance means gathering statistical data maintained by hosts in the managed network. This data tracks internal activity like system CPU and memory utilization, and different call statistics like response time, recognition time, and so on. Analyzing performance data means to make sense of it, see trends, and find ways to improve the performance.

The Management Station periodically collects performance data from managed hosts. The Management Station also regularly pings each host to determine whether the host and its services are running or have generated any alarms. Alarms are sent to the Management Station in real time.

The Management Station displays the performance data as graphs and the operational status as LEDs for an at-a-glance assessment of the overall system health. The display is updated every several seconds giving you a current glimpse of your system as events occur.

Viewing operational status

To view the operational status of managed hosts, follow these steps:

1. Click the Monitoring & Control tab.
2. Click System View. All configured hosts appear in the left frame.
3. Expand the tree view for one of the hosts. All configured services for that host appear.

   Each element in the network tree displays an LED indicating its operational status. Based on the color code, you can quickly determine if the hosts and services are online, if any alarms have been generated, and if so, the severity of the alarm. Nuance defines three alarm levels: minor, major, and critical, which appear in the Management Station as yellow, orange, and red. See Alarms.

   For example:

   

   The LED colors mean:

   • Gray—Configured but not running (offline).
   • Green—Online and no alarms.
   • Yellow—Minor alarms, the system encounters an error that briefly and partially affects a service instance or host.
   • Orange—Major alarms, the system encounters a serious error leading to the loss of a service instance.
   • Red—Critical alarms, the system encounters a catastrophic failure leading to the complete loss of a service or set of services.
   • Nuance Network LED—Shows the status of the most severe alarm that has occurred in the system, including any alarms on the Management Station host (which does not appear in the tree view).

The network tree display refreshes every 10 seconds. To change the timer, see the User Preferences page (see Modify user preferences).

Monitoring vital signs

The Management Station displays performance data on various aspects of the system every several seconds. These statistics, called vital signs, are displayed for the network, cluster, host, and service scopes.

To view the vital signs, follow these steps:

1. Click the Monitoring & Control tab.
2. Click System View.
3. Select the scope from the network tree. (Hosts and services must be online to display vital signs.)
   1. Click Vital Signs on the toolbar. Depending on what you select, you can see:
      • Nuance Network: System CPU and memory-use statistics for all hosts in the network, and audio cache hits for the Nuance Speech Server.
      • Cluster: System CPU and memory-use statistics for all hosts in the cluster.

        Note: The vital signs displayed for the network and cluster scopes are the same. The values reflect the current scope. See Network and cluster vital signs.

      • Host: CPU and memory-use statistics for that host. See Host vital signs.
      • Service: CPU and memory-use statistics for that service. These vital signs are common to all services. The Speech Server display additional vital signs. See Service vital signs.

The Management Station charts the vital signs over a running 15 minute window and updates the display every 30 seconds. For example: