NetBeez enables users to define alerts with several parameters, assign them to Tests, and receive a notification when an alert is triggered. Some metrics available to generate alerts are, for example:
- PING packet loss
- PING round-trip-time
- DNS query time
- HTTP loading time
- Traceroute number of hops
Add/Edit Alert Profiles
- Access the Anomaly Detection Tab in the NetBeez Settings and go to the Alert Profiles section.
- To create a new Alert Profile, click the plus sign in the bottom table row. To edit an existing Alert Profile, click on the corresponding cog icon. A new window will appear, titled Create Alert Profile (or Edit Alert Profile).
- Input a Name and select the Alert Type. The available Alert Types are Up-Down, Performance Baseline, and Performance Watermark.
- Up-down Alerts are triggered when a user-determined number of consecutive tests fail. Up-down Alerts can be applied to ping, DNS, HTTP, or traceroute tests.
To configure an Up-Down test, select Up-Down from the Alert Type select menu.
Input the number of consecutive failed tests to trigger the alert, and select the type of test from ping, DNS, and HTTP.
- Performance Baseline Alerts are based on historical data and moving averages. A performance baseline alert is triggered when a short-term test average deviates from its long-term average. The short-term average can be 1 minute, 15 minutes, 1 hour, or 4 hours while the long-term average can be set to 1 day, 1 week, or 1 month. By comparing the short with the long term average, we can detect ongoing performance degradation issues at a location without having to set a fixed threshold.
For example, if the 15 minute average round-trip time of a ping test is 150 ms while the 1 month average round-trip time is 30 ms, a performance baseline alerts are triggered by setting the multiplier to 5.
To configure a Performance Baseline Alert, select Performance Baseline from the Alert Type select menu.
Select the metric. Metrics supported by Performance Baseline Alerts are Packet Loss (PING), RTT (ping), Failed Test (DNS), Lookup Time (DNS), Failed Test (HTTP), and Response Time (HTTP).
Select the time period over which to measure performance.
Input a multiplier. This multiplier allows for measurement of performance in relational, rather than numerical terms.
Select the comparison period.
- Performance Watermark Alerts are used to enforce Service Level Agreements (SLA). Performance watermark alerts are triggered when a short-term average crosses a user-determined threshold. Also here, the short term average can be set to 1 minute, 15 minutes, 1 hour, or 4 hours.
For example, the user can be alerted if the 15-minute average page loading time of a website is longer than 3 seconds.
To configure a Performance Watermark Alert, select Performance Watermark from the Alert Type select menu.
Select the metric. Metrics supported by Performance Watermark Alerts are Packet Loss (PING), RTT (ping), Failed Test (DNS), Lookup Time (DNS), Failed Test (HTTP), and Response Time (HTTP).
Select the time period over which to measure performance. Input a threshold value. The unit of this value depends on the metric selected.
- Performance Baseline AND Watermark are triggered when both a baseline and a watermark condition are met.
- After configuring your Alert Profile, click Save to exit the Alert Profile window.
- To delete an Alert Profile, click the corresponding delete icon, represented by a red minus sign icon.
In the top left section of the Alert Profiles tab, you'll see a slider button to enable or disable the calculation of percentile-based means.
This function, if enabled, creates for each test two statistics: the mathematical average, which is available by default, and the percentile-based mean, which is enabled with this function. The percentile-based mean is used to filter out outliers that fall outside two times the standard deviation interval (from the mathematical average).
The benefit of this function is that, should a single data point skew the mathematical average, the percentile-based mean won't be affected, thus reducing the number of false positives (alert noise).
To enable the usage of the percentile-based mean on your Alert Profiles, either edit existing alert Profiles or create new Alert Profiles and check the newly available percentile-based mean metric option, as seen in the screenshot below:
This option is available once you switch the percentile-based means calculation "ON" and choose one of the following alert types from the dropdown menu: PING: RTT, DNS: Lookup Time, or HTTP: Response Time.
Attach Alert Profiles to Targets
Performance Alert Profiles can be attached to one or more applicable Targets and Resources.
- Select the Target where you want to apply Alert Profiles, and then click the Edit Target menu entry.
- Locate the Alerts Configuration section at the bottom of the Edit Target panel. Alerts will be shown in a table.
- Select the Alert Profile(s) that you would like applied to the Resource by clicking the checkbox in the column for that Resource. Only applicable Alert Templates will be clickable.
- Click Save.
On Agent Alerts
When the server loses/re-establishes connectivity to an Agent, an alert is triggered. This type of alert can also be sent via email or SNMP trap.
To receive an email notification, the email recipient of the alert has to be added in the Default Emails field under the SMTP Alerts configuration panel (see below).
Configure Alert Notifications
You may choose to be notified via SMTP or SMNP when Performance Alerts are triggered. You may also configure alerts to be sent as a Syslog message. For more details about configuring Alert notifications, please see: