Query Examples
maiLink Telemetry supports the PromQL query syntax. Below is a set of examples of querying in maiLink Telemetry. Apply what you learn here in the maiLink Telemetry Metrics Explorer.
But First …
■ Specifying Devices
In maiLink Telemetry queries, you can use the device= syntax to specify the device context for plotting.
To display data for a specific device use:
deviceID="12345"
where “12345” is the unique Device Identifier for that device.
To display data for the “current” device use, and be able to use the very same query across other devices of the same type, use:
deviceID="$device.id"
■ Beware of Data Overload
The choices you make when querying data can impact performance. For instance, if you make the selected period too long, you may try and retrieve too much data and slow performance. If you make the sample rate too small, you may similarly try and retrieve too much data and slow performance.
■ Beware of Sampling Error
If you try and retreive too little data, you can run into other issues. For instance, if your selected period is too short, you might not observe important events within your data. Or, if you make the sample rate too long, then you might inadvertently wind up with aliasing, where you miss important details in the data.
Smoothing Data
You can easily smooth the data with functions like avg_over_time():
avg_over_time(OvenTemperature__C{device=$device.id})[5m]
In this instance the temperature is averaged over time, creating a rolling average of your temperature. This smooths the data to let you observe more general trends in the data. It gets rid of the “noise” by effectively applying a low-pass filter.
Query Examples
■ Example 1
Suppose you have a been sending a metric called “OvenTemperature__C” to maiLink Metrics. To plot the time series you can use the query:
OvenTemperature__C{device=$device.id}
This will display the raw data from the time series … for data over the selected period, sampled at the selected sample rate. Be aware that there is a balance you must make.
■ Example 2
Suppose you have a been sending a software heartbeat to maiLink SRM once per minute. In theory you should be receiving 60 heartbeats per hour. The raw matric you could plot is:
MySoftwareHeartbeat{device=$device.id}
This should plot as a horizontal line, which isn’t very interesting. Instead, perhaps you might plot the number of heartbeats received per hour. This would allow you to see if the rate ever changes, like taking the derivative of the data. You can do this by summing the data over a period of an hour:
sum_over_time(MySoftwareHeartbeat{device=$device.id})[1h]
Additional Resources
Further information is available at: