Showing posts with label Azure Stream Analytics. Show all posts
Showing posts with label Azure Stream Analytics. Show all posts

Tuesday, May 05, 2015

IOT best practices discussed at the recent Build Conference

This post is written after listening to the video presentation of the talk. This post will be continued.

Best Practices for Creating IoT Solutions with Azure talk delivered by Kevin Miller at the recent build conference covered the Microsoft strategy of dealing with 'IOT' and coming up with some patterns and practices.

The components of IOT, from devices to presentation are as shown in this slide. The whole path from devices to presentation and action is chalked out. In addition to excellent storage for both traditional and non-traditional data there is also a strong backing of Analytics. The demo later in the talk describes it in some detail.

The pattern and practices is the time honored way to move ahead in a rational manner. The idea is not to start something big, but start small and build up. The reason is of course, to make sure the application works with fewer devices connected and then think about scaling it. Of course most of this is common sense. It is not just connectivity but a whole lot of issues like management/security are involved and therefore it is prudent not to start off big and start reducing.

 
Since the IOT covers such numerically large and varied devices; and services, one has to really think about the requirements in the very beginning. Since it is difficult to implement security after the application is built, it is important to include security requirements in the very beginning. These security requirements may span the whole range from devices to presentation at every level. The added complexity that one may encounter with devices is that all devices do not have the same level of security features as those features may depend on the complexity of the device and the device manufacturers approach to security. It is also possible that every device on the 'IOT' do not need the same level or kind of security feature. There is a lot that is business end-use related. In some cases security take a back bench while for others it is the most crucial element.
 
What was not covered was about the automaton objects that industries already has, is there going to be some kind of retro-fitting needed to make them 'IOT Compliant'?
--to be continued
 
 

Tuesday, March 17, 2015

Microsoft Azure IoT Services update

At Convergence 2015 in Atlanta, GA Microsoft outlined Microsoft's vision of IoT on 16th March. Convergence is actually showcasing of Microsoft Dynamics CRM products but as PowrBI is very much a part of IoT a large part of the talk from the CEO was on 'Azure IoT Suite' and how Windows 10 will play out in IoT.

Here is the complete Azure IoT Services suite, some production worthy(almost) and others in preview:
Preview Stage-Likely to enter GA stage next month
  • Azure DocumentDB
  • Azure Stream Analytics
  • Azure Machine Learning
  • Power BI
In general availability(GA)stage
  • Azure Notification Hubs
  • Azure HDInsight
  • Azure Event Hubs
Then there is the 'Azure Intelligent Systems Service' consisting of:
  • PowerBI
  • HD Insight-Microsoft's Big Data Offering via Microsoft's Hadoop which is later expected to debut as a comprehensive  Azure IoT Suite
Here you can find lot more details on most of the above:

DocumentDB
http://hodentek.blogspot.com/2014/08/nosql-azure-service-documentdb-as.html
http://hodentek.blogspot.com/2014/08/how-do-you-sign-up-for-azure-documentdb.html
http://hodentek.blogspot.com/2014/08/getting-to-know-documentdb.html
http://hodentek.blogspot.com/2014/08/getting-to-know-documentdb-part-2.html

Machine Learning
http://hodentek.blogspot.com/2014/07/azure-ml-predictive-analytics-as.html

Hadoop:
http://hodentek.blogspot.com/2014/02/windows-azure-managment-improvements-in.html

PowerBI
http://hodentek.blogspot.com/2015/02/power-bi-unchained.html
http://hodentek.blogspot.com/2015/02/spin-out-cutting-edge-report-with-power.html
http://hodentek.blogspot.com/2015/02/powerbi-preview-reporting-from-sql_13.html
http://hodentek.blogspot.com/2015/02/powerbi-preview-reporting-from-sql.html
http://hodentek.blogspot.com/2015/02/powerbi-preview-reports-using-data-on.html

Azure Event Hubs:
http://hodentek.blogspot.com/2014/11/move-towards-large-scale-azure-adoption.html 

Sunday, November 02, 2014

Real-time analysis of IoT on Microsoft Azure with Stream Analytics

Simply put, it is about collection of data from various devices (different types, different protocols, different complexity, etc) in real-time and aggregating and processing this incoming stream of data to provide useful information. This can enormously improve efficency and make the business stand out. Presently businesses may be implementing custom solutions for addressing such data handling and analysing scenarios.

This is going to get more streamlined, easy and cost-effective using the recently introduced Azure Stream Analytics. Azure Stream analytics has all the well known advantages of Azure Cloud; fully managed; real-time stream computation; highly resilient; and easy to implement and easy to get started. It can, not only help larger corporations who want to dissociate themselves from custom solutions but also help small businesses who cannot afford a custom solution.

All it takes are a few click to author a streaming job using a simplified SQL-like language and monitor the outcome. According to Microsoft, stream rates of kB/sec to gb/sec are accomodated. There are no custom codes to write as most of it is declarative (language).

An assoiciated product, the Azure Event Hub to which Stream Analytics connect to feed all the collected data from varied devices. The streaming nature of data is compromised neither during this 'injesting' process nor during the computational phase of the analytics.

The advertised key capabilities are the following:
  • Ease of use -declarative query model, customer insulation from computautional complexity
  • Scalability -handle millions of events/sec
  • Reliable, repeatable and quick recovery-guarantees zero data loss
  • Low Latency -Optimized for sub-second latency with an adoptive pull-based model
  • Reference data- treated very much like the incomimg stream
The following scenarios of usage are mentioned:
  • Financial Services; Personalized stock trading and alerts
  • Real-time fraud detection
  • Identity protecction real-time
  • Web click stream analytics
  • Telemetry log analysis
  • Event archival for future reference
Here is a schematic of the stream analytic process (with blog author superposed lines in yellow).



Read more here (as this post is a condensed version):
http://azure.microsoft.com/en-us/documentation/articles/stream-analytics-introduction/
DMCA.com Protection Status