kasto - Fotolia
Microsoft executive T.K. "Ranga" Rengarajan kicked off PASS Summit 2014 in Seattle this week with a rundown of upcoming and recent SQL Server advances -- most notably, a planned update to the company's cloud database service that will include more functionality from the on-premises SQL Server software.
The theme of Rengarajan's keynote at the annual conference of the Professional Association for SQL Server (PASS) user group was, as he put it, "data is the thing that is going to light up the future of productivity." He said the number of data-gathering devices is growing and the amount of store data will continue to increase as well. Every year, according to Rengajaran, sees a 40% increase in the data collected by organizations.
To help users deal with all that data, Rengarajan, a corporate vice president who leads engineering for Microsoft's database and big data businesses, announced plans for a major update to the Azure SQL Database cloud service. Due out in a preview release by the end of the year, the update will increase the extent to which SQL Server users can work with the cloud, Rengarajan said; it will also improve T-SQL compatibility and add new parallel querying and database monitoring features. In addition, the updated version of Azure SQL Database will support the use of in-memory OLTP and column-store indexes within a single table. This allows for simultaneous optimization for both analysis and OLTP and, according to Rengajaran, will help integrate the Azure SQL Database workload with on-premises SQL Server databases.
Past, present and future analysis
Rengarajan also promoted the idea that organizations need to use data to "understand the past, analyze the present and predict the future." To meet that objective, Microsoft recently introduced three new cloud applications that were highlighted during the keynote session: Azure Data Factory, Azure Stream Analytics and Azure Machine Learning.
TK 'Ranga' RengajaranMicrosoft executive
Data Factory can take in massive amounts of data and make it accessible for analysis. Sanjay Soni, a senior technical product manager at Microsoft, demonstrated the product by showing how a heat map of customer interest in various in-store displays at retailer Pier 1 Imports can be used to determine what products should be featured on aisle "endcaps." The Stream Analytics tool automates the data analysis process on streams of event data, while Azure Machine Learning is where the analytics become predictive.
To demonstrate the capabilities of Azure Machine Learning, Soni used another example from Pier 1 -- a loyalty program application for the iPhone. The loyalty program tracks past purchases made by a specific customer and by other shoppers who fit a similar profile, then uses the predictive analytics tool to make personalized recommendations in real time as the customer enters the store. As part of the demo, the program predicted that Soni was most likely to be interested in buying beer glasses.
Harnessing the power of prediction
Tiffany Wissner, senior director of data platform marketing at Microsoft, said in a later interview that she thinks predictive analytics "is going to be one of the key differentiators" for businesses. If one company can predict what is coming and another can't, the first has a major advantage, she said.
Eron Kelly, Microsoft's general manager of SQL Server product marketing, added that the software vendor uses Azure Machine Learning for fraud detection and cybersecurity purposes. He also detailed another use case outside of Microsoft. The escalators in the London Underground, he said, are equipped with sensors that monitor the vibration of the escalator wheels. Azure Machine Learning predicts when the escalators will need maintenance based on the vibration frequency of the escalator wheels. When the vibration deviates past a certain point, an alert is sent that preventive maintenance is needed. Kelly described the process as "making decisions based on what is going to happen."
Rengarajan also treated the keynote audience to a sneak peek at a capability currently in development that Microsoft refers to as a "stretch table." This capability would take a SQL Server table and stretch it into Azure SQL Database, so that part of a single database table could be on-premises and part in the cloud. The goal is a seamless user experience: An application looking at a stretched table wouldn't see that it's partially in the cloud and partially on-premises, Rengarajan said.
Wissner described several possible use cases for stretch tables. One example is to keep current processes on-premises and move historical data to the cloud. Others include keeping data that requires extra security on-premises while moving everything else to the cloud, and putting large video files in the cloud while keeping the less resource-intensive elements of a table on-premises. But Microsoft has not announced a release date for the stretch table capability.
There are changes ahead for Azure SQL Server, including new subscription tiers
Gilt uses machine learning to predict what they'll need to ship
Check out this guide to using predictive analytics to answer business questions