Mitel Performance Analytics allows you to configure performance thresholds to generate alarms when the thresholds are crossed.
- Threshold alarms can have the following alarm severities:
- Warning – No immediate impact, but abnormal device behavior detected
- Minor – Non performance impairing
- Major – Performance impairing
- Critical – Device out of service
For each alarm severity level, the system applies value and time hysteresis to reduce the number of spurious alarms.
For example, the system can generate a minor alarm for IP SLA when packet loss is ≥ 2% for at least 10 minutes. The alarm is cleared when packet loss < 2% for at least 5 minutes.
Performance thresholds can be set for the following parameters:
- Probe check-in time
- IP SLA packet loss
- Ping (ICMP) round-trip time
- Ping (ICMP) packet loss
- CPU usage
- Memory usage
- Interface availability
- Bandwidth utilization
- Windows service inactivity
- Voice Quality R value
Windows service thresholds can be set per device by specifying the Windows Service(s) to be monitored.
Interface thresholds (for availability and bandwidth utilization) can be applied to the following interface types:
- ds1- 1.5 Mbps serial interface
- ds3 - 45 Mbps serial interface
- e1 - 2.0 Mbps serial interface
To set performance thresholds for a series of devices, you must first determine which Probe is monitoring those devices.
System performance thresholds are configurable from the container of the Probe that is monitoring the devices; or that container’s parent containers up to the root of the container structure. A system administrator can set system wide-thresholds at the root container of a structure. The system thresholds apply to all Probes in the container structure; and therefore to all devices in the system. A local administrator, who has access to just a few containers with just one Probe for example, can set thresholds for just the containers and the single Probe they can access. The local threshold settings apply to just the devices monitored by the local Probe.
Performance thresholds are applied hierarchically throughout a container structure. A threshold set in a lower container in the hierarchy overrides the same threshold set higher in the hierarchy.
For example, a system administrator can set the following system thresholds for bandwidth utilization:
- Raise a minor alarm when utilization ≥ 75% for longer than 20 minutes; clear the minor alarm if it drops to < 70% for longer than 10 minutes
- Raise a major alarm when utilization ≥ 85% for longer than 20 minutes; clear the major alarm if it drops to < 80% for longer than 10 minutes
The previous thresholds apply to all Probes and monitored devices in the system.
A local administrator who can access a single Probe, called Probe A for example, can set the following thresholds from Probe A’s container:
- Raise a minor alarm when utilization ≥ 65% for longer than 15 minutes; clear the minor alarm if it drops to < 60% for longer than 10 minutes
- Raise a critical alarm when utilization ≥ 95% for longer than 10 minutes; clear the critical alarm if it drops to < 90% for longer than 15 minutes
As a result of the previous settings, Mitel Performance Analytics generates the following alarms for the devices monitored by Probe A:
- Raise a minor alarm when utilization ≥ 65% for longer than 15 minutes; clear the minor alarm if it drops to < 60% for longer than 10 minutes (set locally)
- Raise a major alarm when utilization ≥ 85% for longer than 20 minutes; clear the major alarm if it drops to < 80% for longer than 10 minutes (inherited from the system thresholds)
- Raise a critical alarm when utilization ≥ 95% for longer than 10 minutes; clear the critical alarm if it drops to < 90% for longer than 15 minutes (set locally)
Only users with the appropriate privileges can set thresholds
to configure system thresholds, do the following steps:
1. Determine which Probe is monitoring the devices you want to set thresholds for.
2. Access the Probe’s container dashboard or the dashboard of a parent container. Select Thresholds under the System Administration menu.
The Global Thresholds window is displayed, showing a table of all parameters with thresholds for all device types. The parameters are listed in the left. The device types make up the table columns. The following is an example:
Note: The Global Thresholds window displays settings that apply to the current container and all its descendant containers only. It does not display settings inherited from parent or other ancestor containers.
Each element in the table indicates the most severe threshold that has been configured and enabled for that combination of parameter and device type. The following table describes the threshold icons that appear in the Global Thresholds window.
Thresholding is not supported.
The threshold is supported but not defined.
The threshold is defined but not enabled.
Most severe alarm defined for this threshold is Warning.
Most severe alarm defined for this threshold is Minor.
Most severe alarm defined for this threshold is Major.
Most severe alarm defined for this threshold is Critica
Hovering over a threshold icon provides more details on the thresholds defined for that particular parameter and device combination.
In the following example, there is one threshold alarm configured for MiVoice Business Voice Quality.
For further information on the threshold alarms, click the icon at the intersection of the parameter and device type. The Thresholds page is displayed.
The Thresholds page shows the thresholds settings for that parameter and device pair. In this example, a minor alarm is generated when the R value is ≤ 70 and cleared when it increases to > 70 for 10 minutes.
Click the Edit button on the Thresholds page to open the Edit page.
The threshold Edit page allows you to set warning, minor, major and critical alarm severity threshold values for a specific parameter and device pair.
Some parameters and thresholds pairs can have a Match value required for the threshold to be valid. The parameters with Match values are:
- Interface Availability, RX Bandwidth Utilization and TX Bandwidth Utilization: Matches are selected from: ppp, pppMultilinkBundle, propPointToPointSerial, hdlc, sdlc, e1, ds1, or ds3.
- Disk Usage: Matches are entered as text strings corresponding to an OS volume or mount point name. The following are examples:
- For Windows: C:\, or E:\data
- For Linux and Unix: /, /root, or /data
- Services Inactivity: Matches are entered as text strings to the Windows Service name field. The Windows Service name is available from the Windows Service Management console. The following is an example to monitor for the availability of PostgreSQL running on a Windows Server. In this example, you identify the Windows Service name using the Services Console.
Double-click on the Postgres service to find the Service Name (not the Display Name).
From the information on the previous dialog fox, you enter postgresql-9.0 as the match parameter.
Lower is Worse?
Separate thresholds can be set for parameters which indicate worse conditions with higher values and for parameters which indicate better conditions with higher values. The default settings assume that a higher parameter value indicates a worse condition, except for the R value used in the Voice Quality thresholds.
For example, to generate an alarm if CPU usage is less than a certain value, ensure that the Lower is Worse? checkbox is selected.
You can set multiple alarm severity thresholds for each combination of parameter, device, match and lower is better.
Set threshold values by dragging the slider for the selected alarm severity or by entering the value below the slider. The slider moves to the nearest valid values for the threshold it is being set for. For instance, if the Ping Packet Loss is measured in 25% increments, the slider jumps to the nearest 25% increment.
Time and Value Hysteresis
Mitel Performance Analytics supports both time and value hysteresis for both raising alarms and clearing alarms. This feature reduces the number of nuisance alarms.
For example, in the previous graphic with standard settings for device CPU and memory utilization threshold alarms:
- CPU utilization:
- A minor alarm is raised at 9:45 (utilization ≥ 75% at 9:30 + 15 minutes on time requirement).
- A major alarm is raised at 10:15 (utilization ≥ 85% at 10:00 + 15 minutes on time requirement).
- A major is downgraded to a minor alarm 10:55 (utilization < 80% at 10:45 + 10 minutes off time requirement).
- A minor alarm is cleared at 11:25 (utilization < 70% at 11:15 + 10 minutes off time).
- Memory utilization
- A minor alarm is raised at 10:30 (utilization ≥ 85% at 10:15 + 15 minutes on time requirement).
- A minor alarm is cleared at 11:10 (utilization < 80% at 11:00 + 10 minutes off time).
The time and value hysteresis criteria are combined. In the previous example, if the memory utilization increases momentarily to 75% at 11:20, the alarm is not cleared until 11:30 (11:20 + 10 minutes off time requirement).