2016, the Year of Internet of Things
Welcome to 2016 you lovely Cloudy Readers. Hope you managed to take a few days’ rest and eat some [relevant seasonal food]. Ready to get back into it? ‘Course you are. "Internet of Things" (IoT) looks like it will be to 2016 what "Big Data" was to 2015 in terms of level of hype and sales waffle.
Perhaps even more worryingly, as the concept also links to vast quantities of information, it provides further opportunities for cheesy terms like "single pane of glass" to continue to be banded about. However if one cuts through this nonsense, as we like to do here at Cloudy Towers, useful opportunities appear providing the information you need to make significant and value adding business decisions.
What is IoT?
Just in case you’ve been living on a desert island for a few years…. Basically it is when "sensors" or "chips" are built into physical devices allowing the bi-directional flow of information. As a spurious example, that could be a chip in your kettle sending information to your phone about temperature and permitting remote control of the unit. On a more useful level, perhaps it could be a sensor built into an industrial press which logs data 24/7 and 365 on every aspect of its operation, and feeds this back to a central storage location for machine learning driven analysis of performance.
While collecting sensor data is nothing new perhaps, cloud based IoT brings with it:
- *massive* scale potential, seamlessly integrated with data analytics pipelines to generate insight from the data collected
- the opportunity to implement event-driven remote control over the "things"
- standardisation of SDKs and connectivity protocols designed to be power efficient and handle intermittent connectivity
Why should I care?
Well, running with the "kettle" theme, there are actually useful examples in the home. One of the more commonly cited examples is the Google owned Nest range of thermostats. Not only can this thermostat be controlled remotely, but using sensors it can learn your behaviour patterns and adapt to your "usage" of your property. Keeping you at the right temperature, and saving you money. Phillips’ Hue lightbulb system is also interesting, although it has already ignited a debate regarding standardisationin this brave new IoT world.
Generally, this is going to be *big* for consumers. It’s going to be expected, and demanded. It’s going to make your life easier as an individual. It might even keep you alive longer.
To give this some context regarding scale, a recent Gartner report I saw stated the following:
"Approximately 3.9 billion connected things were in use in 2014 and this figure is expected to rise to 25 billion by 2020."
Crave a monetary value to help gauge this? Ok, well McKinsey documented the potential economic impact of IoT at $11.1 trillion per year by 2025. By 2025, the Yuan will probably be our international currency of choice, but that’s a whole different blog post…
Yes, yes, but what about my business?
This is where things get even more interesting. There’s a use case for every business. To get your creative juices flowing, here are a few examples:
- How about water companies measuring rainfall flow and forecasting flood alerts with increasing accuracy?
- Gathering data and changing behaviour around hand-washing in hospitals?
- Logistics companies with perishable goods (potentially anything from vaccines to cabbages) monitoring temperature and reacting before spoilage occurs
- How about predictive maintenance? Whether that’s on a lift, a vending machine, a school bus or anything else requiring an engineer, if you can gather data about the state of a "thing" prior to a fault across million of things or operations, you can predict when it may fail and do something about it *beforehand*
- Smart cement, which sends data regarding stresses and strains and allows architectural failure to be spotted before it occurs
- Streamlining the efficiency of waste collection by understanding how full bins are
- Lastly of course, a "thing" could also be an animal. It could be a human, or even a cow
If it exists, you can gather data from it, and learn as a result.
What can the major cloud vendors offer?
The most important point here is "scale". Think about it. While sensor data is small (tiny in fact) in isolation, big streams would overwhelm most businesses at scale. These volumes of data would typically be construed as a DDOS attack in any normal IT environment…
The good news is that both Microsoft and Amazon launched services back in March 2015 relating to IoT. Microsoft released theirs just a week in advance of AWS, which shows you how competitive things are between the major players.
Both support the HTTP and MQTT protocols (if you don’t know what the latter is, you will in a year) and platforms from vendors like Intel and Texas Instruments. Of course, both allow integration to other proprietary platform services, e.g. Kinesis for AWS or Azure Machine Learning. There are some comparisons here in a nice table, and here in a bit more depth with accompanying text.
Neither offering is 100% finished yet, but both are delivering value already and will evolve and iterate quickly in 2016 based on your feedback. Importantly, they do make the process of creating the data collection, storage and processing pipeline simpler, allowing you to focus on gaining value from the data.
The biggest challenge here, as with any new area of technology, is a lack of skills. There aren’t that many people who understand "big data" and "IoT" in any depth. This isn’t a huge problem if you approach projects with smart people, and external assistance where needed, but it’s worth bearing in mind.
Also note that, depending on your business, you may need to process data locally via edge devices of some form, so you can react to data changes quickly. This may need to happen in milliseconds in some cases, i.e. the potential value gleaned may be determined by how quickly you can analyse the data. Sending back to a single central location (cloud or otherwise) may be too slow.
There is a security risk here. Anyone who says otherwise doesn’t understand how to defend a network. In a world where everything is interconnected, you have instantly widened the attack vectors. Jeep found this out in the summer with an entertainment system being compromised, but allowing control of somewhat more important aspects of a vehicle. Like the brakes….
A clearly defined, and programmatically enforced security policy, combined with proactive threat management can of course mitigate this.
Finally, we come to the practical point of battery life – something which blights most of our daily lives today. If your "things" won’t have mains power, you will need to think about how to keep them running with replacement batteries at frequent intervals. Things are improving in this space, see here for an interesting read on IoT designed chipset platforms coming in the future which could last for up to 10 years.
So, should we be undertaking an IoT project in 2016?
Yes. There’s a huge opportunity here. Try it and see. Don’t be afraid to fail, to learn, and then to do it again more effectively. AWS have an effectively priced (cheap!) starter kit available.
Do work closely with your business units to get the best results: domain knowledge of the data being collected is mandatory – this is not just an "IT" project.
As we enter 2017, let’s make sure we’ve helped to close the IoT skills gap.