Azure Stream Analytics is a fully managed PaaS service that helps you run real-time analytics and complex event processing logic on telemetry from devices and applications. Numerous built-in functions available in Stream Analytics helps users build real-time applications using simple SQL language with utmost ease. By using these capabilities customers can quickly realize powerful applications for scenarios such as fleet monitoring, connected cars, mobile asset tracking, geofence monitoring, ridesharing, etc.
Today, we are excited to announce several enhancements to geospatial features. These features will help customers manage a much larger set of mobile assets and vehicle fleet easily, accurately, and more contextually than previously possible. These capabilities are available both in the cloud and on Azure IoT edge.
Here is a quick run-down of the new capabilities:
Previously, to track ‘n’ number of assets in streaming data across ‘m’ number of geofence reference data points, in the geospatial context, translated into a cross join of every reference data entry with every streaming event thus resulting in an O(n*m) operation. This presented scale issues in scenarios where customers need to manage thousands of assets across hundreds of sites.
To address this limitation, Stream Analytics now supports indexing geospatial data in relevant queries. When indexed, geospatial data is joined with streaming events. Instead of generating a cross join of every streaming event with reference data, an index is created with the reference data of geospatial objects and every lookup is optimized using the index. This will enable a faster reference data lookup to O(n * log m), thereby offering support for scale that is magnitudes of order higher than what was previously possible.
GeoJSON is an open standard format designed for representing simple geographical features, along with their non-spatial attributes, based on JavaScript Object Notation. Previously, Azure Stream Analytics, did not extend support for all the types otherwise defined in GeoJSON specification. As a result, users could not successfully export some of their geospatial objects and process them in Stream Analytics.
To remedy this gap, we are adding full support for WKT geospatial format in Stream Analytics. This format is natively supported by Microsoft SQL Server and hence can be readily used in reference data to represent specific geospatial entities or attributes. This will enable users to easily export their data into WKT and add each entry as nvarchar(max).
Previously, in Stream Analytics we implemented Geographical calculations without the possibility of geometric projections. This would mean that users would ingress projected coordinates and expect calculations to follow geometric projections. Unfortunately, in many cases the output would not match their expectation as calculations were based on geography and were ignoring projections.
To help users overcome this limitation and to allow full fidelity projected calculations, we are moving away from geographic based computation and towards geometric calculations. This means that developers can now input their projected geo coordinates using the same functions as before, but the output will preserve their projection properties. That said, ST_DISTANCE function will continue to be the only function over geography.
Here is a quick run-down of the new capabilities:
Geospatial indexing
Previously, to track ‘n’ number of assets in streaming data across ‘m’ number of geofence reference data points, in the geospatial context, translated into a cross join of every reference data entry with every streaming event thus resulting in an O(n*m) operation. This presented scale issues in scenarios where customers need to manage thousands of assets across hundreds of sites.
To address this limitation, Stream Analytics now supports indexing geospatial data in relevant queries. When indexed, geospatial data is joined with streaming events. Instead of generating a cross join of every streaming event with reference data, an index is created with the reference data of geospatial objects and every lookup is optimized using the index. This will enable a faster reference data lookup to O(n * log m), thereby offering support for scale that is magnitudes of order higher than what was previously possible.
Support for WKT format
GeoJSON is an open standard format designed for representing simple geographical features, along with their non-spatial attributes, based on JavaScript Object Notation. Previously, Azure Stream Analytics, did not extend support for all the types otherwise defined in GeoJSON specification. As a result, users could not successfully export some of their geospatial objects and process them in Stream Analytics.
To remedy this gap, we are adding full support for WKT geospatial format in Stream Analytics. This format is natively supported by Microsoft SQL Server and hence can be readily used in reference data to represent specific geospatial entities or attributes. This will enable users to easily export their data into WKT and add each entry as nvarchar(max).
Geometry based calculations
Previously, in Stream Analytics we implemented Geographical calculations without the possibility of geometric projections. This would mean that users would ingress projected coordinates and expect calculations to follow geometric projections. Unfortunately, in many cases the output would not match their expectation as calculations were based on geography and were ignoring projections.
To help users overcome this limitation and to allow full fidelity projected calculations, we are moving away from geographic based computation and towards geometric calculations. This means that developers can now input their projected geo coordinates using the same functions as before, but the output will preserve their projection properties. That said, ST_DISTANCE function will continue to be the only function over geography.
Posts
0 comments:
Post a Comment