GE Addresses Time Series Data Demands

When many think of General Electric (GE), the first thought that comes to mind might not be software, but the company’s unit that focuses on providing code solutions is, according to GE, part of a $4 million business on its own.

GE Intelligent Platforms, the segment of GE that provides software and system elements for automation and embedded computing for industry, military, aerospace and communication system customers, is at the root of that booming software core. This week the division announced new advances to support big data for customers in the data-heavy verticals GE supplies.

The software package, called Proficy Historian 4.5 gathers, aggregates, organizes and manages large datasets in real time for large customers who doubtlessly have rather complex data management demands, including those in the industries mentioned above. According to GE, this software can “improve visible, provide context to raw data, and aggregate islands of information” to allow for more robust, timely decision-making.

GE Intelligent Platforms claims that the new software offering provides support for more than 15 million tags across a scalable platform, which can be integrated without a massive overhaul and provides the built-in ability to allow data sharing across the enterprise or research organization. The data collection element can collect details from a large number of different devices and sensors in real-time.

To put the new offering in context, GE pointed to a use case this week, which involved its own GE Monitoring and Diagnostics Center, which used the package to monitor over one thousand gas turbines around the world in real-time. This monitoring effort required high performance on time-series data.

The team behind the turbine monitoring project had turned to GE to find new architectures, hardware and software to accommodate the flood of data. They found that the Proficy solution “was shown to excel against relational and unstructured databases for time series data.” GE also said that the typical column-based and relational databases that had been used for this purpose were unable to deal with time series process data queries at scale, and that they felt some of the unnamed “technologies” designed to do so with this type of data were still too immature.