What Is GE Predix Really Building?
Author: Dan Woods
While GE’s Predix platform has been promoted on TV and covered in the New York Times and other leading publications, none of the press coverage has explained concretely what technology GE is building with Predix. This story will do just that based on my visit to the recent Predix Transform conference, interviews with people building applications using Predix, and work done creating content to explain Predix for GE.
The standard definition of GE’s Predix is that it is a platform for creating applications for the Industrial Internet of Things. The New York Times puts it this way:
G.E. reimagined Predix as a cloud-based operating system for industrial applications.
But, as my friend Pano Anthos, an veteran entrepreneur who now runs XRC Labs often says, “What does that mean?”
The bouncy and intelligent young engineers on TV commercials suggest that large industrial organizations that use jet engines, trains, and other massive stuff worth billions that create billions more in value will somehow be able to do new and exciting things because of Predix. This much is true. But how? What is the concrete machine that Predix is building so that all of this can happen?
Predix Creates Transformation by Extending Industrial Automation to the Cloud
The first thing to understand about Predix is that it is the next phase of a decades long evolution of industrial automation. The beginning of this journey was about computer control of machinery and industrial processes. Systems described with acronyms like SCADA (Supervisory Control And Data Acquisition) ruled the day and the goal was to increase the efficiency and performance of technology directly controlling high value industrial equipment.
The prize that Predix is after is system wide optimization. Instead of making one piece of equipment better, Predix aims to create a detailed model that spans the entire system. The view created by this model allows both improved optimization of each part of the system and never before possible optimization of the entire system. When a company gets this right, the result is not just an incremental improvement but a transformation to a new level of efficiency and performance.
The foundation of Predix is GE’s long history in industrial automation. GE has business units that are focused on almost every area of industry from aircraft to power plants, manufacturing, transportation, and so on.
GE and other companies have built the systems to control and run the heavy equipment. When a turbine weighing a ton is whirring around at a high speed, the systems controlling it had better be well engineered. Let’s use the word “edge” for all of the systems that directly control industrial equipment.
Predix is about building a bridge from the edge to the cloud. On a locomotive, for example, GE has a system that acts as something like cruise control for the engineer driving the train. That engineer makes many decisions that have a huge economic and safety impact. Cruise control helps make better decisions.
What Predix does for a locomotive is enrich its cruise control with something like Waze, but focused on far more complex systems than commuter traffic. Now each train can benefit from what happened on every other train, including rail, environment, weather and engine performance data and factor in that information. This works because information is gathered from the edge systems of the entire fleet of locomotives and sent to the Predix cloud, which can find insights in the information and send data and suggestions back.
In addition, more sensors can be added to the train to send more information such as video from cameras that show the quality of the track and so on. Sending this information to the cloud allows more intelligence to be added to the system. If one locomotive sees track that is bad, it can send a message to an application running on the Predix cloud that can tell all the other locomotives to adjust their plans to avoid going too fast.
In addition, the information describing everything that is known about the locomotive can be sent to a Predix application that creates a “digital twin” of that particular locomotive. A digital twin is a computer model of an individual physical asset. This digital twin can be use to further analyze the performance of the equipment and in many other ways as well. Because the digital twin exists in the cloud, it can be analyzed based on the system-wide view that Predix assembles, huge data sets that are collected from all other assets, and advanced analytics that are not possible to run at the edge. For a locomotive, a huge stream of data about every aspect of the train’s operation flows back to the Predix cloud. The digital twins can be used to predict failures, optimize performance, find ways the model doesn’t match the physical asset so the digital twin can be improved, and find patterns of both failure and success.
The roots of the transformation Predix seeks to achieve can be found in the interplay and feedback loops between the edge systems enhanced by Predix, which can be seen as remote data centers, and the models and analytics in the central Predix cloud, which see the big picture.
GE believes Predix is a crucial next step that super charges industrial automation to achieve transformation to new levels of performance. Making an industry more efficient by a fraction of a percentage can mean a return of hundred of millions or billions of dollars. Achieving a 5 percent improvement can mean a return in the billions.
Main point: Predix adds a brain with a massive memory and compute power to edge systems for digital industrial transformation.
Every Cloud Is Not the Same
It is important to remember that Predix is not about creating the edge systems, which must be highly engineered, exhaustively tested, and intensely secured. The Stuxnet virus shows the kind of damage that can be done when one of these controllers is hacked.
Predix however will be communicating with the edge systems. For that reason, Predix applications must run in a new kind of cloud, one with much higher service level agreements for security and performance than the first generation of clouds have. Industrial processes are often highly regulated and require documentation and proof of reliability and procedures for managing failures. Predix is built to withstand such scrutiny.
If you look at the fine print for Amazon and other public clouds, you will find very weak SLAs. The reliability, performance, and security of public clouds have improved markedly since their wobbly beginnings but still, the reliability is not sufficient for industrial applications (see “Why GE is Tackling a Harder Problem than Amazon”).
Main point: Predix seeks to create a cloud with the needed security and high performance to safely interact with the edge systems and critical infrastructure.
Adopting the Cloud Foundry Platform
Predix needed a framework that allowed for implementation of a secure high performance cloud in a way that could be easily extended with many services. Cloud Foundry, an open source framework for creating cloud platforms, provides all the software plumbing to enable construction of the Predix cloud.
Cloud Foundry provides a structure for creating a set of services that can interact with each other. In essence, Cloud Foundry provides a sort of grammar for implementing a cloud. Predix uses that grammar but adds value by providing the nouns and verbs to create industrial applications by creating services.
One reason it is important for Predix to define a set of services is to allow the implementation to change as technology gets better. All of the latest stuff that make developers drool such as graph databases, high performance caches, NoSQL, and the like are powering Predix services. For example, Predix provides a time series service that allows streams of data with timestamps to be stored and analyzed. Predix implements this one way today, but in two years a better way may appear. At that time, the underlying implementation may change but the definition of the service will stay the same, ensuring that developers do not have to rewrite applications developed using Predix. The goal is to make these services future proof, so innovations can be added in an orderly manner.
Main point: Predix uses Cloud Foundry to create future-proof services to build industrial applications.
What Are Some of the Predix Services?
Read full article: forbes.com/sites/danwoods/2016/09/28/what-is-ge-predix-really-building/