SEATTLE – Connected devices tend to generate a lot of data, and extracting useful information from that data requires analysis, typically in the cloud. But getting that data ready for analysis can be the lion's share of the extraction effort as well as time consuming. Analytics service provider Glassbeam has launched new services and software that the company claims can automate and speed data preparation as well as move analytics closer to the edge for real-time information extraction.
“There's a chronic problem in the data world,” Glassbeam's founder Punit Pandit told EE Times in an interview. “Today's complex machines are constantly generating all kinds of data, but in different formats. Take an MRI machine, for instance. It has numerous sub components all generating information on the settings, health, and status of the subsystem, and not necessarily in a coordinated fashion. So you end up with multilayer log data. As a result, 60% to 80% of the time it takes to perform analytics on that data is spend on data preparation, transforming, and cleaning in order to get that data into a single database.”
Glassbeam, which provides data analytics services to companies such as IBM and Gridscape, has taken steps to reduce that effort. It has introduced Glassbeam Studio, a cloud-based GUI tool for transforming and preparing unstructured machine log data for analysis. Using a drag-and-drop interface, customers can create a data transformation script in SPL (semiotic parsing language) that will parse, store, and index unstructured log data into a structured format suitable for analysis. Out-of-the-box analytic capabilities include functions for trending, charting graphing, and the like.
Once the transformation script is ready, data preparation becomes automatic, freeing analysts to build models and make predictions from the data. Creating such a script through conventional programming can take months, Pandit noted, while Glassbeam studio reduces the effort to mere days.