Tutorial: Create and configure a Time Series Insights Gen2 environment

In this tutorial, you learn how to create and configure an Azure Time Series Insights Gen2 environment to integrate with your IoT Plug and Play solution. Use Time Series Insights to collect, process, store, query, and visualize time series data at the scale of Internet of Things (IoT).

In this tutorial, you

  • Provision a Time Series Insights environment and connect your IoT hub as a streaming event source.
  • Work through model synchronization to author your Time Series Model.
  • Use the Digital Twins Definition Language (DTDL) sample model files that you used for the temperature controller and thermostat devices.

Note

This integration between Time Series Insights and IoT Plug and Play is in preview. The way that DTDL device models map to the Time Series Insights Time Series Model might change.

Prerequisites

Before you continue, make sure you've set up your environment, including your IoT hub.

At this point, you have:

  • An Azure IoT hub.
  • A Device Provisioning Service (DPS) instance linked to your IoT hub. The DPS instance should have an individual device enrollment for your IoT Plug and Play device.
  • A connection to your IoT hub from either a single-component device or a multiple-component device that streams simulated data.

Prepare your event source

The IoT hub you created previously will be your Time Series Insights environment's event source.

Important

Disable any existing IoT Hub routes. There's a known issue with using an IoT hub with routing configured. Temporarily disable any routing endpoints. When your IoT hub is connected to Time Series Insights, you can enable routing endpoints again.

On your IoT hub, create a unique consumer group for Time Series Insights to consume. In the following example, replace my-pnp-hub with the name of the IoT hub you used previously.

az iot hub consumer-group create --hub-name my-pnp-hub --name tsi-consumer-group

Choose a Time Series ID

When you provision your Time Series Insights environment, you need to select a Time Series ID. It's important to select the appropriate Time Series ID. This property is immutable and can't be changed after it's set. A Time Series ID is like a database partition key. The Time Series ID acts as the primary key for your Time Series Model. For more information, see Best practices for choosing a Time Series ID.

As an IoT Plug and Play user, for your Time Series ID, specify a composite key that consists of iothub-connection-device-id and dt-subject. The IoT hub adds these system properties that contain your IoT Plug and Play device ID and your device component names, respectively.

Even if your IoT Plug and Play device models don't currently use components, you should include dt-subject as part of a composite key so that you can use components in the future. Because your Time Series ID is immutable, Microsoft recommends enabling this option in case you need it in the future.

Note

The examples in this article are for the multiple-component TemperatureController device. But the concepts are the same for the no-component Thermostat device.

Provision your Time Series Insights environment

This section describes how to provision your Azure Time Series Insights Gen2 environment.

Run the following command to:

  • Create an Azure storage account for your environment's cold store. This account is designed for long-term retention and analytics for historical data.
    • In your code, replace mytsicoldstore with a unique name for your cold storage account.
  • Create an Azure Time Series Insights Gen2 environment. The environment will be created with warm storage that has a retention period of seven days. The cold storage account will be attached for infinite retention.
    • In your code, replace my-tsi-env with a unique name for your Time Series Insights environment.
    • In your code, replace my-pnp-resourcegroup with the name of the resource group you used during setup.
    • Your Time Series ID property is iothub-connection-device-id, dt-subject.
storage=mytsicoldstore
rg=my-pnp-resourcegroup
az storage account create -g $rg -n $storage --https-only
key=$(az storage account keys list -g $rg -n $storage --query [0].value --output tsv)
az tsi environment gen2 create --name "my-tsi-env" --location eastus2 --resource-group $rg --sku name="L1" capacity=1 --time-series-id-properties name=iothub-connection-device-id type=String --time-series-id-properties name=dt-subject type=String --warm-store-configuration data-retention=P7D --storage-configuration account-name=$storage management-key=$key

Connect your IoT Hub event source. Replace my-pnp-resourcegroup, my-pnp-hub, and my-tsi-env with the values you chose. The following command references the consumer group for Time Series Insights that you created previously:

rg=my-pnp-resourcegroup
iothub=my-pnp-hub
env=my-tsi-env
es_resource_id=$(az iot hub create -g $rg -n $iothub --query id --output tsv)
shared_access_key=$(az iot hub policy list -g $rg --hub-name $iothub --query "[?keyName=='service'].primaryKey" --output tsv)
az tsi event-source iothub create --event-source-name iot-hub-event-source --environment-name $env --resource-group $rg --location eastus2 --consumer-group-name tsi-consumer-group --key-name iothubowner --shared-access-key $shared_access_key --event-source-resource-id $es_resource_id --iot-hub-name $iothub

In the Azure portal, go to your resource group, and then select your new Time Series Insights environment. Go to the Time Series Insights Explorer URL shown in the instance overview:

Screenshot of the portal overview page.

In the Explorer, you see three instances:

  • <your pnp device ID>, thermostat1
  • <your pnp device ID>, thermostat2
  • <your pnp device ID>, null

Note

The third tag represents telemetry from the TemperatureController itself, such as the working set of device memory. Because this is a top-level property, the value for the component name is null. In a later step, you make this name more user-friendly.

Screenshot showing three instances in the Explorer.

Configure model translation

Next, you translate your DTDL device model to the asset model in Azure Time Series Insights. In Time Series Insights, the Time Series Model is a semantic modeling tool for data contextualization. The model has three core components:

Define your types

You can begin ingesting data into Azure Time Series Insights Gen2 without having predefined a model. When telemetry arrives, Time Series Insights attempts to automatically resolve time series instances based on your Time Series ID property values. All instances are assigned the default type. You need to manually create a new type to correctly categorize your instances.

The following details outline the simplest method to synchronize your device DTDL models with your Time Series Model types:

  • Your digital twin model identifier becomes your type ID.
  • The type name can be either the model name or the display name.
  • The model description becomes the type's description.
  • At least one type variable is created for each telemetry that has a numeric schema.
    • Only numeric data types can be used for variables, but if a value is sent as another type that can be converted, "0" for example, you can use a conversion function such as toDouble.
  • The variable name can be either the telemetry name or the display name.
  • When you define the Time Series Expression variable, refer to the telemetry's name on the wire and to the telemetry's data type.
DTDL JSON Time Series Model type JSON Example value
@id id dtmi:com:example:TemperatureController;1
displayName name Temperature Controller
description description Device with two thermostats and remote reboot.
contents (array) variables (object) See the following example.

Screenshot showing D T D L to Time Series Model type.

Note

This example shows three variables, but each type can have up to 100 variables. Different variables can reference the same telemetry value to do different calculations as needed. For the full list of filters, aggregates, and scalar functions, see Time Series Insights Gen2 Time Series Expression syntax.

Open a text editor and save the following JSON to your local drive.

{
  "put": [
    {
      "id": "dtmi:com:example:TemperatureController;1",
      "name": "Temperature Controller",
      "description": "Device with two thermostats and remote reboot.",
      "variables": {
        "workingSet": {
          "kind": "numeric",
          "value": {
            "tsx": "coalesce($event.workingSet.Long, toLong($event.workingSet.Double))"
          },
          "aggregation": {
            "tsx": "avg($value)"
          }
        },
        "temperature": {
          "kind": "numeric",
          "value": {
            "tsx": "coalesce($event.temperature.Long, toLong($event.temperature.Double))"
          },
          "aggregation": {
            "tsx": "avg($value)"
          }
        },
        "eventCount": {
          "kind": "aggregate",
          "aggregation": {
            "tsx": "count()"
          }
        }
      }
    }
  ]
}

In the Time Series Insights Explorer, select the model icon on the left to open the Model tab. Select Types and then select Upload JSON:

Screenshot showing how to upload JSON.

Select Choose file, select the JSON you saved previously, and then select Upload.

You see the newly defined Temperature Controller type.

Create a hierarchy

Create a hierarchy to organize the tags under their TemperatureController parent. The following simple example has a single level, but IoT solutions commonly have many levels of nesting to contextualize tags within their physical and semantic position within an organization.

Select Hierarchies and then select Add hierarchy. For the name, enter Device Fleet. Create one level called Device Name. Then select Save.

Screenshot showing how to add a hierarchy.

Assign your instances to the correct type

Next you change the type of your instances and place them in the hierarchy.

Select the Instances tab. Find the instance that represents the device's working set, and then select the Edit icon on the far right.

Screenshot showing how to edit an instance.

Open the Type drop-down menu and then select Temperature Controller. Enter defaultComponent, <your device name> to update the name of the instance that represents all top-level tags associated with your device.

Screenshot showing how to change an instance type.

Before you select Save, first select the Instance Fields tab, and then select Device Fleet. To group the telemetry together, enter <your device name> - Temp Controller. Then select Save.

Screenshot showing how to assign an instance to a hierarchy

Repeat the previous steps to assign your thermostat tags the correct type and hierarchy.

View your data

Go back to the charting pane and expand Device Fleet > your device. Select thermostat1, select the Temperature variable, and then select Add to chart the value. Do the same for thermostat2 and the defaultComponent workingSet value.

Screenshot showing how to change the instance type for thermostat2.

Clean up resources

If you plan to continue with additional IoT Plug and Play articles, you can keep and reuse the resources you used in this article. Otherwise, you can delete the resources you created in this article to avoid additional charges.

You can delete both the hub and registered device at once by deleting the entire resource group with the following Azure CLI command. Don't use this command if these resources are sharing a resource group with other resources you want to keep.

az group delete --name <YourResourceGroupName>

To delete just the IoT hub, run the following command using Azure CLI:

az iot hub delete --name <YourIoTHubName>

To delete just the device identity you registered with your IoT hub, run the following command using Azure CLI:

az iot hub device-identity delete --hub-name <YourIoTHubName> --device-id <YourDeviceID>

You may also want to remove the cloned sample files from your development machine.

Next steps

To learn more about the various charting options, including interval sizing and y-axis controls, see Azure Time Series Insights Explorer.