Alarm Policies

You can set alarm policies for metrics and events of a cloud service. When a metric triggers the threshold in the alarm policy for multiple times in a specified period, you will be notified. This section describes how to configure alarm policies for metrics and events.

Configuring Alarm Policies for Metrics

You can monitor key metrics of cloud services by configuring alarm rules. Then you can handle exceptions in a timely manner. A metric alarm policy must include a metric name, statistic, consecutive triggering times, threshold, and frequency. For details, see the following table.

Items in an alarm policy for metrics

Item	Description	Example Value
Metric Name	Metric name. After selecting a metric, move the cursor to on the right of the metric to view its description. For details about the monitoring metrics supported by cloud products, see Cloud Product Metrics.	CPU Usage
Statistic	Metric value type. Cloud Eye supports the following statistics for metrics: Raw data, Avg., Max., Min., Variance, and Sum. Raw data indicates the metric data that is not processed or converted. Avg. is the value calculated by averaging raw data during a rollup period. Max. is the highest value observed during a rollup period. Min. is the lowest value observed during a rollup period. Variance: indicates the difference between each data point in the original value and the average value within a rollup period. Sum is the sum of raw data during a rollup period. NOTE: An aggregation period can be 5 minutes, 20 minutes, 1 hour, 4 hours, or 24 hours. Select an aggregation period based on your service requirements. If you set a rollup period, alarm notifications will be delayed. If you set the rollup period to 5 minutes, alarm notifications will be delayed for 10 to 15 minutes. If you set the rollup period to 20 minutes, alarm notifications will be delayed for 20 minutes. If you set the rollup period to 1 hour, alarm notifications will be delayed for 1 hour and 20 minutes. If you set the rollup period to 4 hours, alarm notifications will be delayed for 4 hours and 40 minutes. If you set the rollup period to 24 hours, alarm notifications will be delayed for 25 hours. When selecting a rollup period, make sure it's longer than the metric reporting period. Otherwise, alarms may not generated. For example, if the metric reporting period is 10 minutes, you cannot create an alarm policy with a 5-minute rollup period.	Raw data
Consecutive Triggering Times	Number of consecutive times that an alarm is triggered. The value can be set to 1, 2, 3, 4, 5, 10, 15, 30, 60, 90, 120, or 180 times (consecutively).	2 times (consecutively)
Operator	Operator used to compare metric value and the threshold. Cloud Eye supports >, >=, <, <=, =, !=, Increase compared with last period, Decrease compared with last period, and Increase or decrease compared with last period. NOTE: Increase compared with last period: The metric data reported in the current monitoring period increases sharply when compared with that in the previous monitoring period. Decrease compared with last period: The metric data reported in the current monitoring period decreases sharply when compared with that in the previous monitoring period. Increase or decrease compared with last period: The metric data in the current monitoring period increases or decreases sharply when compared with that in the previous monitoring period.	=
Threshold and Alarm Severity	Threshold for triggering an alarm and alarm severity.	Critical: 22 Byte/s
Frequency	How often alarms are repeatedly notified when there is already an alarm. The following options are available: Trigger only one alarm, Every 5 minutes, Every 10 minutes, Every 15 minutes, Every 30 minutes, Every 1 hour, Every 3 hours, Every 6 hours, Every 12 hours, and One day. NOTE: If the alarm is not cleared after it is generated, an alarm notification is sent, once every hour.	Every 5 minutes

Example of configuring an alarm policy for a metric

For example, in an alarm policy, the metric name is CPU usage, the statistic is average, the rollup period is 5 minutes, the consecutive triggering times is 2, the operator is =, the threshold is 80% (critical), and the frequency is every 5 minutes.

This alarm policy indicates that the average CPU usage is collected every 5 minutes. If the CPU usage of an ECS is greater than 80% for two consecutive times, a critical alarm is generated every 5 minutes.

Figure 1 Alarm policy for a metric
Click to enlarge

Configuring Alarm Policies for Events

You can configure alarm policies for various system and custom events so that you can take measures in a timely manner when an event occurs. An event alarm policy must include the event name, triggering period, triggering type, triggering times, and alarm frequency. For details, see the following table.

Items in an alarm policy for events

Item	Description	Example Value
Event Name	Name of a service event.	Startup failure
Triggering Period	Event triggering period. The following options are available: Within 5 minutes, Within 20 minutes, Within 1 hours, Within 4 hours, and Within 24 hours. NOTE: This parameter is optional when you select Accumulative trigger.	Within 5 minutes
Trigger type	The value can be: Immediate trigger (default): After the event occurs, an alarm is triggered immediately. Cumulative trigger: An alarm is generated only after the event is triggered for a preset number of times within the triggering period.	Accumulative trigger
Triggering times	Cumulative number of times the event occurred within the triggering period. NOTE: This parameter is optional when you select Accumulative trigger.	2
Frequency	How often alarms are repeatedly notified when there is already an alarm. The following options are available: Trigger only one alarm, Every 5 minutes, Every 10 minutes, Every 15 minutes, Every 30 minutes, Every 1 hour, Every 3 hours, Every 6 hours, Every 12 hours, and One day. NOTE: This parameter is optional when you select Accumulative trigger.	Every 5 minutes
Alarm Severity	Alarm severity, which can be Critical, Major, Minor, or Informational.	Major

Example of configuring an alarm policy for an event

For example, in an alarm policy, the event name is startup failure, the triggering period is 5 minutes, the trigger type is cumulative trigger, the triggering times is 2, the alarm frequency is once every 5 minutes, and the alarm severity is major.

This alarm policy indicates that a major alarm is generated every 5 minutes if the startup failure event is triggered for 2 consecutive times within 5 minutes.

Figure 2 Alarm policy for an event
Click to enlarge