The need for such infrastructure is critical: the GEnx jet engine, for example, can produce some 5,000 data points per second to optimise flight times. Harnessing, analysing and synthesising that much data will require powerful and robust modalities, according to Piero Bonissone, a Coolidge Fellow who has written about the “Three Vs” of Big Data.
Volume: whereas 15 years ago, a typical PC carried 10 gb of total storage, the GEnex jet engine could well produce up to 10 tb of data about its performance in only half an hour.
Velocity: Machine-to-Machine (M2M) processes share concurrent, real-time data among billions of sensor-embedded devices—devices that could number 30bn by 2020.
Variety: Big Data could also be called Wide Data, 3D Data, Geospatial Data or Unstructured Data; all not readily handled by the kind of traditional databases used by expensive, finite single servers.
As a result of the overwhelming stream of data being unleashed, larger companies are looking to cloud technology for storage and retrieval. Single servers can corrupt or lose data, are hard to scale up and only can be accessed by limited numbers and types of machines. Cloud platforms offer all of these and more. The cloud comes at a cost, and smaller businesses may find the current average $100 per user per month price point untenable. Larger enterprises, however, are jumping in, with much competition among cloud service providers. (Amazon has a wide lead, although many others also are in play.)
Large multinationals are outfitting trains, appliances, power meters and turbines with cloud-ready sensors, with smaller players also getting in on the action. California-based Ayla Networks, for example, has created an integrated “plug-and-play” solution to make devices and appliances intelligent at production, thus allowing them to access the Industrial Internet’s cloud immediately. Norway’s ThinFilm prints rewritable sensors on labels thinner than a human hair; smart tags can be attached to perishable foods to alert food vendors about temperature and freshness.
Such smart tags can send data either directly to the cloud or to the cloud after a processing stop at mobile devices. Indeed, the proliferation of mobile technology goes hand-in-hand with the rise of the cloud. As they gain more powerful capabilities, mobile devices increasingly act as bridges (and local computing hubs) between embedded machines and the cloud. On a consumer level, such interlinking is already unlocking new product potential—such as a bed that monitors sleep and raises a snorer’s head or remotely diagnosing issues with heavy farm equipment.
Cloud platforms also bring together disparate data groups: GE’s Pivotal and Predictivity services give airlines, hospitals and railroads a common architecture for operating and managing critical machinery. Such platforms are also enticing cities, Leesburg, Florida, for one, to install and operate cloud-based smart grids for their municipal utilities as they look to save millions in power costs over the next few decades.
As Big Data cloud storage and analytics spread, possibilities open up for the next big trend for the Industrial Internet—cloud-based industrial apps similar to those found in consumer app stores for smartphones and tablets but structured to improve equipment efficiency. Consumer app producers rely on high-volume sales of inexpensive products and don’t necessarily make much money.
Industrial apps, though, have the potential to deliver significant financial value to end-users, create lasting ecosystems around platforms (i.e. individual cloud providers) and increase productivity. Industrial app “stores” and other infrastructure remain to be built, but the benefits of productivity-based industrial apps could leave both app producers and their happy clients floating, possibly on cloud nine.
This piece first appeared on GE Look ahead.