IBM Patents Real-Time Analytics for Cloud Data
IBM has patented a technique for real-time analysis of cloud data as well as a technique for using analytics to optimize cloud computing performance.IBM announced it has patented an invention that enhances the use of analytics for assessing and directing data in a cloud computing environment, enabling more timely and efficient application processing and management. The IBM invention, U.S Patent #8,639,809, "Predictive Removal of Runtime Data Using Attribute Characterizing," analyzes data from a variety of sources to avoid performance lags and processing delays. Since not all data is equal and resources for processing, storing and managing information are finite, real-time analytics can be useful in expediting this process, IBM said. "Processing data in a cloud is similar to managing checkout lines at a store—if you have one simple item to purchase, an express lane is preferable to waiting in line behind someone with a more complicated order," IBM Inventor Michael Branson, who co-invented the patented technique with John Santosuosso, said in a statement. "Cloud customers don't want data that can be analyzed and dealt with simply to sit idle behind data that needs more complex analysis. Applying real-time analytics in a cloud can help ensure each piece of data gets the proper attention in a timely manner." IBM's newly patented technique performs real-time analysis on data—such as online transactions, readings from sensors, financial quotes and video streams—as it is generated. It quickly determines how each piece of data will be processed, by identifying patterns in the data values that have correlated with slower processing in the past, to avoid situations where certain data values might delay the processing of all the data.
The cloud computing system then automatically channels each piece of data—also known as a tuple—down the right path for timely analysis. IBM explains that the invention creates an "express lane" for "normal" tuples, ensuring that they are analyzed promptly, while tuples with values known to be problematic or that are very time-consuming to analyze are sent down other analysis paths.