The Texas Instruments LaunchPad sends telemetry including simulated temperature and humidity readings, alternatively telemetry can be produced via an Azure Web Site webpage using a fake telemetry generator. The telemetry is then ingested by Azure Event Hub and processed by Azure Stream Analytics in near real time, the processed telemetry is then stored in an Azure SQL Server Table. One table stores telemetry that has just been processed, where as another table stores averaged temperature and humidity data, along with the data count for a 5 minute tumbling window.
Alerts can be generated via a Stream Analytics query that compares values stored in a SQL Database Table, these alert details are sent to Event Hub and can be used to generate an SMS, email or phone alert using an Azure API App and an External System like Twilio. Alternatively notification can be sent to Mobile Apps and be viewed as a Smart Watch Alert.
Live data can be viewed in a moving graph as Stream Analytics send processed data to Power BI. Newly stored processed telemetry can also be viewed as it arrives via an Azure Web App webpage as can averages for time periods.
A predictive model is created using Azure Machine Learning and made available as a web service. The model can be queried from an Azure Website webpage to determine whether a system requires maintenance. Azure Stream Analytics can also query the predictive models via the Azure Machine Learning Web Service and hence provide near real-time predictions for many different scenarios.