Government // Enterprise Architecture
Commentary
1/11/2012
02:56 PM
Chris Murphy
Chris Murphy
Commentary
Connect Directly
Google+
LinkedIn
Twitter
RSS
E-Mail
50%
50%

GE's Huge Software Ambition: 3 Key Takeaways

GE picks Silicon Valley as the best place to innovate a new generation of software aimed at connecting industrial machines.

Ruh sees a number of factors coming together to spur growth of the industrial Internet now:

• Lower-cost sensors that can collect data from more machines.

• Lower-cost GPU processors that let a device sift through data and send back only the bits that are out of the norm and in need of further analysis. "You don't want every piece of data going back to the cloud," Ruh says.

• Analytics and management capabilities advancing to the point that they can make sense of the data coming back from devices.

• Computing infrastructure that's cheap and flexible enough to make it practical to process this stream of data.

For CIOs, the takeaway is that machine-to-machine data connections, backed by analytics and automated response, are getting more practical.

3. The industrial Internet needs a lot of work

It's a lot different from today's consumer Internet. For example, industrial endpoints (machines) will create massive amounts of data, while the consumer Internet is more about human endpoints consuming data. Analytics to predict outcomes like a machine breakdown is essential, as is automation to deal with the data volume.

The Silicon Valley center--under construction in San Ramon and due to open around June--will develop what Ruh calls a "unifying architecture" for all of GE's industry software. GE and others are in the early days of understanding what's needed to drive an industrial Internet. "That's why it'll take 10 years," Ruh says, for the ecosystem to develop.

This is emerging tech, so the gains will come in pockets, with early adopters reaping the pain and the gain. Customers will want to see measurable returns on this investment--remote monitoring systems that show they can reduce maintenance staff headcount, reduce machine downtime and thus increase utilization.

Previous
2 of 2
Next
Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
ChrisMurphy
50%
50%
ChrisMurphy,
User Rank: Author
1/30/2012 | 10:20:24 PM
re: GE's Huge Software Ambition: 3 Key Takeaways
Your point's a very good one -- they wouldn't want to ignore data they'll later need for analysis. The larger point is that we can start doing more processing at the point of collection, on the edges of networks, and we'll need to when we get into these big data extremes that machines can produce. One simple illustration is an example that i link to above, with AccuWeather. It does processing at the edge of its network (through its content delivery network vendor) to truncate GPS data when a smartphone asks for a weather forecast. If it doesn't do that, the data center thinks two requests 100 yards apart are different, churning through CPUs to get a new forecast, when in fact the weather outlook's exactly the same in both places. With a few lines of code, AccuWeather can drastically cut the amount of processing it needs to do, and the data flowing back and forth over the network.
ANewNickname
50%
50%
ANewNickname,
User Rank: Apprentice
1/19/2012 | 10:34:07 AM
re: GE's Huge Software Ambition: 3 Key Takeaways
I've got a GE home security system, but GE has apparently abandoned this area. The company which bought this line seems to be new to the world of software and the concept of rapid response to user needs. This is a ripe area for open-source development.
kkrugler
50%
50%
kkrugler,
User Rank: Apprentice
1/18/2012 | 2:49:28 PM
re: GE's Huge Software Ambition: 3 Key Takeaways
The point about "...send back only the bits that are out of the norm..." struck me as odd. When you're using time-series data to predict future problems, you don't know yet what's a good predictor, so you don't know what bits to send vs. toss. And if you view the prediction of problems as something similar to fraud detection, then you can't do much sampling of the data, as the key signals are random and scattered.

Also, if you view a factory/power plant/whatever as something similar to a complex software system, where outputs from one component are inputs to another, then it also becomes hard to know in isolation (on a single device) what bits matter when trying to build a learning model for the system.

So from the description above, it sounds like GE is focusing on the much more narrow problem of embedding detectors in devices that "phone home" when they're pretty sure something is out of whack.
Jim C, Orlando
50%
50%
Jim C, Orlando,
User Rank: Apprentice
1/17/2012 | 5:31:31 PM
re: GE's Huge Software Ambition: 3 Key Takeaways
Actually, the consumer market already has GM's OnStar and home burglar alarm monitoring.
The analytics could start out pretty basic similar to any dashboard -- you just have to figure out the green, yellow and red zones. The statistics for figuring out in control vs. out of controll industrial conditions (control charts) were pioneered by Shewart and Deming almost a century ago. With GE's engineering background and six sigma expertise, they should have a pretty good idea about at what value equipment is out of specifications and at what value it is dangerously out of spec.
ChrisMurphy
50%
50%
ChrisMurphy,
User Rank: Author
1/12/2012 | 11:01:23 PM
re: GE's Huge Software Ambition: 3 Key Takeaways
Agree, and you can include GE on this list of companies that have been doing this for years. But it's about where this is headed. The pace of change the next 5 years will be a lot faster in this arena than the past 5 years. Wireless capabilities, cheaper sensors, cheaper point-of-collection processing, better analytics -- those forces will open up entirely new options for collecting industrial data and doing predictive analytics than have been possible. In 2006, you could've said Salesforce has been doing cloud computing for years, what's the big deal? But it was in the next five years that the explosion of new uses of cloud apps (and infrastructure and development platforms) really took off.
bblair495
50%
50%
bblair495,
User Rank: Apprentice
1/12/2012 | 7:16:36 PM
re: GE's Huge Software Ambition: 3 Key Takeaways
What am I missing? Companies such as Infor, IBM, Rockwell and even small players like ShopLogix have been doing this for years?
Register for InformationWeek Newsletters
White Papers
Current Issue
InformationWeek Tech Digest, Dec. 9, 2014
Apps will make or break the tablet as a work device, but don't shortchange critical factors related to hardware, security, peripherals, and integration.
Video
Slideshows
Twitter Feed
InformationWeek Radio
Archived InformationWeek Radio
Join us for a roundup of the top stories on InformationWeek.com for the week of December 14, 2014. Be here for the show and for the incredible Friday Afternoon Conversation that runs beside the program.
Sponsored Live Streaming Video
Everything You've Been Told About Mobility Is Wrong
Attend this video symposium with Sean Wisdom, Global Director of Mobility Solutions, and learn about how you can harness powerful new products to mobilize your business potential.