Internet of Things – a view from 33,000ft
I’m on a long haul flight to meet some customers in the US. Disclosure – I work mostly on Pivotal [part of the EMC federation], which remains a hot topic in the market.
I’m surrounded by devices.
Just in my bag I have a laptop, iphone (personal), iphone (work), ipad and a good camera that is wifi enabled – oh, and a tennis swing sensor. So 6 devices, each capable of being connected.
From my seat I can see at least 14 other devices, the most interesting are the currently the relatively simple worn devices – Fitbits and Nike fitness bands. Some can even share with each other peer to peer.
So probably between 400 and 1000 devices just on this aircraft – just in the metal tube that holds the passengers. The fitness devices are quietly collating the data of the snoozing passengers.
On the plane itself we have 4 lovely, reliable engines with 100s or even 1000s of sensors. Airframe, speed sensors, temperature, pitot tubes, radar, fuel levels etc. – More than 4 terabytes of data will be generated by this flight alone.
So what are we seeing?
Setting aside the metal tube filled with snoozing people and gadgets, this theme is all connected to that handy and abused tag line the Internet of Things – covering a lot of emerging behavioral changes.
I split what we are seeing into two domains of interest:

  1. The connected industrial “things”– where every industrial machine and all its sensors, telemetry and life history are connected – that provide both enhanced immediate decision making to improve machine performance – collating back to a manufacturer’s Business Data Lake and user to provide enhanced services to their user base.
  2. The analytical consumer – devices, once passive and dumb taking on whole new capabilities as the data types that can be collected by devices – Fitbit et al – sleep patterns, motion, pulse, golf swing, racket swing, etc. can be interrogated by an app on your phone through the creation of a personal data lake – a data fabric of your life events and activities.

I don’t discount others, I just think these are of pertinent interest right now as we start to discover the real potential of combining big, fast data with analytics. But before we get to the point of being able to get value from the data, we need to be able to ingest, and we have to consider physics:

  • Latency – from sensor capture to a capability to analyze. It sounds trivial but some decisions must be made inside the sensor itself – even if the only decision is to capture or not capture the data. Latency to respond to changing conditions can be critical.
  • Bandwidth – from the sensor to the device to the storage fabric. Don’t assume you can keep everything. You want to keep what you need and, impossibly, you do don’t know what you might be analyzing in the future – so ensure your system is extensible.
  • Connectivity – devices cannot always be connected and even when they can, backhauling data is slow, compared to a Fedex parcel with a hard drive.
  • Storage – building adaptive sensors that can not only make their local decision, but then also know which data should be stored and then promulgated is key. Enriching the data lake of information is key, but not each and every sensor point can be captured – pragmatism is needed.
  • Power – balancing power needs is increasingly important – how do we balance consumer expectations with device capability?

What does this mean to The Internet of Things?

  • Physics is going to get in the way for some decision making and you need to reflect that in your approach


  • Local embedded analytics is key, with decisions and data abstracted to the data lake to allow medium and long term decision making to be performed.

All this to consider, it makes me tired. Time to be another snoozing passenger surrounded by gadgets.