GE Industrial Internet of Things Vision

Last week at GE’s Minds+Machines conference in San Francisco, Jeff Immelt, CEO of GE, spoke about his vision for the Internet of Things. He made several interesting points. First, he said that GE is a software capable company that already generates $6B in software sales annually. While relatively small in comparison to revenue generated by Microsoft, Oracle and SAP, it’s still big enough to rank GE in the top 6 largest software vendors worldwide. He also said that industrial companies will need to become software and analytics companies. Third, he said that IT professionals will be the biggest driver of productivity going forward for industrial businesses. My take of Mr. Immelt’s vision is that GE’s software group isn’t just a strategic, yet relatively minor, revenue-producing initiative. Instead, it will ultimately be a primary way in which GE transforms its own business and the businesses of its customers.

Predix Platform

GE’s industrial transformation strategy is founded on its new Predix platform for the Industrial Internet of Things (IIoT). The platform is a suite of technology and business services (see figure 1) for creating end-to-end IIoT analytics and control applications. Frankly, GE is well positioned to design a platform like Predix because it has decades of experience deploying and managing complex machinery in a wide variety of use cases. Now Predix is not the only IIoT platform in the market. IBM, PTC, SAP, Siemens, Microsoft and many other companies offer or intend to offer similar platforms. But from my viewpoint, Predix offers several interesting and potentially unique services that differentiate it in the marketplace.

Technical and Business Services

While analytics are an essential service within Predix, it also includes other complimentary services that are necessary to assuring a successful IIoT application. These enabling services cover both technical and business factors. For example Predix Connectivity services provide a virtual network that can overcome reliability issues, bandwidth limitations or availability constraints. The network can leverage multiple communication paths including cellular, fixed line, and satellite telecom, using partners that include AT&T, SoftBank, Verizon and Vodaphone. Predix also includes a device/gateway/controller system platform called Machine services. Machine services provides a broad suite of functionality to ensure that edge nodes in an IIoT system operate effectively in their role to capture, manage, secure and communicate device status and control data. Developers gain sandboxes, tools, documentation and component libraries via Predix.io to accelerate development of applications. To secure effective integration with other business offerings, Predix offers BizOps services that manage use payments for deployed systems. GE’s desire to offer their customers a comprehensive suite of services to develop, deploy and monetize IIoT applications will go a long way to assuring success.

Analytics, Data and Metadata Services

The heart of the Predix platform is its ability to analyze IIoT device data. Support for analytics within Predix comes from 3 services; Predix Analytic services, Predix Data services and Predix Asset services:

• Predix analytics in their current form are highly flexible but requires coding competency. Analytics are coded in Java, Matlab/Octave or Python. Analytics are modular and managed in a catalog. With this modular approach, both descriptive and predictive analytics are supported. Each analytic can be visualized using simple web page views or in custom applications.
• Predix Data services ingest, clean, manage and deliver a wide variety of data types including time series, BLOB and tabular/relational data. Ingestion can be streaming (e.g. HTTPS) or batch (e.g. FTP). Predix’s data management includes internally developed time series and BLOG data stores, PostgreSQL, Raik CS and Redis (so far). Message queuing is provided by Rabbit MQ. Pivotal is a key data management partner and received a $105M investment from GE in April 2013.
• Supporting Predix Analytics and Data services are Asset services that define and relate objects to represent a IIoT device network. The device network is represented in a graph model managed by Cassandra. Attribute metadata of nodes on the graph are managed through an associated classification scheme.

Platform Foundation

Predix is based upon Cloud Foundry, an open source Platform-as-a-Service (PaaS) from Pivotal software. Much of the Predix Data services come with this platform.

Cloud Hosting

Production deployment of Predix analytics will be on GE data centers designed to specifically suit the needs of IIoT. (Beta versions of Predix are currently on AWS.) In particular there will be more, smaller data centers that are closer to its customers’ sites. These sites will be independent of GE’s own IT fabric, within sovereign boundaries as required by local regulations and able to use private communication lines to assure data security and privacy. Many of these sites will be located in Equinix data centers.

Take-aways

GE Software Predix is certainly more than just analytics. It is a comprehensive vision for what’s needed to deploy an IIoT application. However with the current version, development and use of analytics by non-programmers is unaddressed, which may inhibit rapid prototyping in some organizations.  I suspect that this functionality will likely be added in future releases. For organizations that wish to evaluate quick IIoT application prototypes to establish concept feasibility, other solutions might be faster to use including GroveStreams, ZoomData and OpenDataSoft. Though for organizations that are intent upon deploying IIoT applications that are revenue producing or mission critical, GE Software Predix should be on the short list.