I recently saw a very interesting presentation, - From Cloud to Fog: The Internet of Things, - by Cisco executive Michael Enescu. He tells us that the Internet of Things (IoT) is truly taking off, already comprising about 25 billion smart devices at the edges of the Internet, a number that’s expected to more than double over the next decade, and continue to keep growing for the foreseeable future.
These devices generate massive amounts of data, a lot of which requires real-time actions. It’s impractical to move all that data to a central cloud for analysis and actions. Computing and intelligence thus have to move closer to the edge, to both ameliorate the data transport challenges and enable real-time actions as required. This all leads to what has come to be known as edge computing, - an architectural bridge between the clouds at the center and the IoT devices all around the edges of the Internet. Cisco calls it fog computing, an allusion to what happens to clouds as they get closer to the ground.
Cisco defines fog computing as “a highly virtualized platform that provides compute, storage, and networking services between end devices and traditional Cloud Computing Data Centers, typically, but not exclusively located at the edge of network.” Fog complements cloud with a number of important capabilities, including edge location awareness, low latency in highly constrained connectivity environments, bandwidth and energy optimizations, and nearly unlimited scale.
Enescu notes that cloud and fog represent two very different IoT paradigms for dealing with huge amounts of data. In cloud, the data generated by the smart sensors at the edges is transferred and stored in the center, where it’s analyzed, and the appropriate actions then flow back across the network. With fog on the other hand, the analysis is done in real time at or near the edge devices, from which actions then flow across the network, with only the data to be stored being transferred to the central cloud.
Fog or Edge computing might well become a new computing model, the 4th such model in the IT industry’s 60 year history. In its early decades, just about all computing was centralized, mostly large and mid-size mainframes and supercomputers. A limited number of users had access to those expensive machines, perhaps a million or two in total.
The 1980s saw the advent of increasingly powerful and inexpensive microprocessors, personal computers and workstations, which paved the way for the transition to the new, distributed client-server model of computing. These systems were optimized for simplicity and flexibility, and their graphical user interfaces led to many innovative applications, including the new web-based applications in the 1990s. The total number of users rose to 100s of millions.
Over time, many companies ended up with large numbers of relatively small, under-utilized servers distributed across the organization. These factors eventually led to significantly increased management complexities and costs. In the mid-late 1990s, the pendulum started to swing back toward a more centralized computing model. Many of the previously distributed servers were consolidated in data centers.
The mid-late 2000s saw the emergence of the Internet-based cloud computing model. The centrally managed, highly scalable, attractively priced cloud platforms were now being designed to support a variety of mobile devices, - e.g. smartphones, tablets, e-readers and laptops. The total number of users is now into the billions.
But, just as we’re getting comfortable with clouds, the IoT has reached an inflection point, with 10s of billions of smart devices of all sorts in the relatively near term, rising to 100s of billions in the decades ahead. The centralized cloud architectures must now be complemented with the highly distributed fog/edge architectures.
This was also the opinion of an excellent paper, Device Democracy: Saving the Future of the Internet of Things, by IBM’s Paul Brody and Veena Pureswaran. They argue that we’re at the threshold of a major economic transformation as a result of the 100s of billions of IoT devices expected over time. Their overall conclusion is that “a low-cost, private-by-design democracy of devices will emerge that will enable new digital economies and create new value, while offering consumers and enterprises fundamentally better products and user experiences.”
This transformation is driven by the continuing advances of technology. Not that long ago, IoT solutions were mostly based on special purpose computers, often including just the minimum necessary to control the devices. “Thanks to Moore’s Law, it is now cheaper and easier to make a device with a powerful general purpose computer than it is to create a customized embedded device. In the fast approaching future, devices from doorknobs to light bulbs will be able to carry as much compute power and connectivity as the first smartphones.”
As a result, the present Internet, which may well become known as the Internet of People, has to now scale from supporting billions of users and mobile devices, to supporting 100s of billions of IoT devices. Doing so requires a major Internet reboot to help it overcome a number of key challenges:
The cost of connectivity. “Even as revenues fail to meet expectations, costs are prohibitively high. Many existing IoT solutions are expensive because of the high infrastructure and maintenance costs associated with centralized clouds and large server farms, in addition to the service costs of middlemen.”
The Internet after trust. “The Internet was originally built on trust. In the post-Snowden era, it is evident that trust in the Internet is over. The notion of IoT solutions built as centralized systems with trusted partners is now something of a fantasy.”
Not future-proof. “While many companies are quick to enter the market for smart, connected devices, they have yet to discover that it is very hard to exit. While consumers replace smartphones and PCs every 18 to 36 months, the expectation is for door locks, LED bulbs and other basic pieces of infrastructure to last for years, even decades, without needing replacement.”
A lack of functional value. “Many IoT solutions today suffer from a lack of meaningful value creation. The value proposition of many connected devices has been that they are connected - but simply enabling connectivity does not make a device smarter or better. Connectivity and intelligence are a means to a better product and experience, not an end.”
Broken business models. “Most IoT business models also hinge on the use of analytics to sell user data or targeted advertising. These expectations are also unrealistic. Both advertising and marketing data are affected by the unique quality of markets in information: the marginal cost of additional capacity (advertising) or incremental supply (user data) is zero.”
Overly optimistic forecasts: “Products like toasters and door locks worked without apps and service contracts before the digital era… many smart device manufacturers have improbable expectations of ecosystem opportunities. While it makes interesting conversation for a smart TV to speak to the toaster, such solutions get cumbersome quickly and nobody has emerged successful in controlling and monetizing the entire IoT ecosystem.”
To overcome these major challenges we must fundamentally rethink the IoT’s technology strategies, business models and overall designs. Supporting 100s of billions of IoT devices at radically lower costs requires a decentralized peer-to-peer architecture, capable of near unlimited scalability, privacy, security and the ability to handle trustless transactions. “In other words, the IoT represents a case of billions of players, not all of which can be trusted - some even malicious - with a need for some form of validation and consensus. And for this, the blockchain offers a very elegant solution… By confirming transactions using decentralized consensus, the blockchain eliminates the need for trust.”
The paper offers a number of IoT business model guidelines, from unlocking the excess capacity of physical assets to digitally integrating value chains. “In the end, the IoT is expected to make the physical world every bit as easy to search, utilize and engage with as the virtual world. We describe this emerging transformation as the liquification of the physical world. Just as large financial marketplaces create liquidity in securities, currencies and cash, the IoT can liquify whole industries, squeezing greater productivity and profitability out of them than anyone ever imagined possible.”
Finally, Brody and Pureswaran remind us that the connectivity and intelligence of 100s of billions of IoT devices in not an end in itself, but a means to create better products and user experiences that calls for good design thinking. “To most users of these devices though, networking and remote usage are only secondary features. Consumers care most about the primary functional value and user experience… Putting the user first, and designing for user-centric experiences and value will be critical to adoption of the IoT.”