Azure Stream Analytics, an Event Processing Engine

Azure Stream Analytics is Microsoft’s event processing engine that allows to run a real-time analytics on multiple IoT or non-IoT data streams. The data can come from sensors, devices, social media feeds, websites, infrastructure systems, applications, and more.

 

 

Azure Stream analytics can be used:

 

  • to monitor high volumes of data streaming from certain processes or devices
  • to extract information from a particular data stream
  • to determine trends, patterns, and relationships

 

Users can use these patterns to activate different processes and actions, for example, automation workflows and alerts, send information to a reporting tool or store this information for the later investigation. Examples, where Stream Analytics can be used, include web clickstream analytics, data protection and fraud detection, stock trading analysis etc.

 

Azure Stream Analytics can get data from IoT Hub, Azure Event Hubs, and Azure Blob Storage. Stream Analytics is also available on Azure IoT Edge. Users need to create Stream Analytics job that will describe exactly the source of data and a transformation using the SQL-like query language to filter, classify, combine into a group, and connect streaming data for a certain period of time. Eventually, the job will specify the output for the transformed data. Users can get actionable insights in almost real time and use them for predictions.

 

The output can be written to a number of storages, for example, Azure Data Lake Stores,   Azure SQL Database, Azure Blob, and Azure Cosmos DB. Besides, Azure Stream Analytics supports Power BI and Azure Service Bus. Stream Analytics is easy to use because, thanks to the simplicity of Stream analytics query language, users can create complex analyses without any programming. But capabilities of the query language can be extended by defining additional functions either in Azure Machine Learning or creating custom code with user-defined functions in JavaScript. Azure Stream Analytics can deal with 1 GB of incoming data per second. The service has built-in recovery features.