Round up of IIoT analytics news

A few stories that caught my eye in the last week or so, including the end of the world as we know it...

• I found this blog, Planning for IoT Analytics Success, to be a pretty good summary of the ecosystem needed for Industrial IoT analytics.  Edge analytics, event streaming, machine learning, time-series data - it's pretty much all touched on.  A nice summary.


• I'm going to bend this next article just a little to suit my own purposes...I've written and talked a lot about machine learning lately.  It's going to be a key analytics technology for IIoT.  Some naysayers (you know who you are...) fret that industrial data is too difficult for unaided algorithms to make sense of.  One of the conclusions from Why machine learning is not the silver-bullet solution to online fraud is that machine learning algorithms still need to be wielded by smart humans to be effective against fraud.  And the same is true in the industrial sector.  Algorithms can be a great help - they can even help solve problems that couldn't really be solved before.  But, those smart algorithms need to be tempered with domain expertise to be really effective.


• IBM (an ARC client) continues its relentless assault on Industrial IoT (IBM Launches Quarks Open Source Development Tool To Build Efficient IoT Apps).  Effectively supporting applications development is one of the 8 key components that Greg Gorbach and I have identified for enterprise IIoT platforms.  In the grand scheme of things, this has been a pretty low-key announcement from IBM.  As in, I can't find any kind of press release for it....Regardless of that, you can learn more about Quarks on IBM's website here.


• I've not been very bullish on Hadoop - at least not as far as the industrial sector is concerned.  I think Hadoop is still some way from crossing the chasm in manufacturing, oil and gas etc.  That said, Hadoop vendor Hortonworks triples revenue in Q4 has to be a positive sign for the Hadoop ecosystem.  That said, Q4 revenue for Hortonworks was still just $37m  - with an annual operating loss on the wrong side of $50m.  So, on the right track perhaps, but not out of the woods yet by any stretch.


• And finally, to pick up on the machine learning thread again, check out the latest promo video from Boston Dynamics (a company now owned by Google).  I find that video both impressive and scary in equal measure.  But, at least we now know who'll be first up against the wall after the robot uprising.  It's the guy with the hockey stick...