Michael Hay

Will Cloud Computing Survive the Internet of Things? (Reflections and Conclusions)

Blog Post created by Michael Hay Employee on Aug 5, 2014

Reflections

Are Cloud Computing and the Internet of Things forces to be reckoned with?  Could cases exist where the Internet of Things makes Cloud Computing obsolete?  This series answered the first question by providing an accounting of Cloud Computing and the Internet of Things.  In the case of Cloud Computing compelling points are exemplified in the scale of various measures of the movement.  Notably Microsoft’s estimated power consumption being equivalent to a small city, and the scale of data production larger than all of the movies ever filmed illustrate that Cloud Computing is indeed a force. Even though the Internet of Things is heavily evolving one can feel the weight of IoT everyday.  From turning on a light bulb on the beach to unlocking mysteries about the Universe IoT is here already and growing.  Perhaps a better view of the velocity of the IoT movement lies in the availability of wearable technologies for fitness and flexible mobile devices like the iPhone.  Actually, Cloud Computing and the Internet of Things have high market velocity and pervasively appear in everyone’s daily life.  Are both forces?  “Force” roughly means energy as an attribute of movement, and by all cited measures, of Cloud Computing and the Internet of Things, market growth or movement are apparent.  Therefore, I assert that both are forces.  And the second question:  Could cases exist where the Internet of Things makes Cloud Computing obsolete?  The answer to this question is more complex, but it begins with Carr’s thesis about Cloud Computing being akin to power utilities.  Perhaps a way to reveal an answer to this question will emerge through an exploration of physical architectures and futures of power production and distribution utilities.


energy.pngRelevant Developments in the US Power Grid

In the United States the power industry began in 1882 with Thomas Edison’s electrification of a portion of New York City [32].  As an industry that has persisted for over 130 years, it has been through and witnessed significant changes.  Today the architecture of the power grid is made up of largely centralized generation, transformation, transmission networks, transmission substations, distribution networks and, finally the load of electricity consumers [32].  Industry and academic reports show that investments for transmission and distribution in particular are anemic [32] [33].

 

Even as demand has skyrocketed, there has been chronic underinvestment in getting energy where it needs to go through transmission and distribution, further limiting grid efficiency and reliability. While hundreds of thousands of high-voltage transmission lines course throughout the United States, only 668 additional miles of interstate transmission have been built since 2000. As a result, system constraints worsen at a time when outages and power quality issues are estimated to cost American business more than $100 billion on average each year [33].

 

According to the US Department of Energy, the solution to underinvestment in transmission and distribution, is not additional investment in long haul power networks but transformation to a smart grid system, “…that is less centralized and more consumer-interactive [33].”  It seems that companies like Apple, Facebook and Microsoft have recognized this point and are investing on their own, supporting exactly this pattern.  These companies appear to be developing a mixture of local production and super energy efficient data centers, and cleverly positioning themselves within the power grid closer to power generation [34] [35] [36].  Therefore, it seems that the power industry, at least in the US, is moving in the opposite direction from mass centralization.  The direction appears to combine centralized production with distributed production and smarter networks via the smart grid.  One instance is Apple’s proposed new campus in Cupertino, CA, where they intend to use the power grid that distributes power from traditional centralized sources as “…a backup [34].”  Overall moves to invest in smart grids with an assumption of anemic investments in the distribution network may prove relevant for the combination of Cloud Computing and IoT.


Conclusions

Actually, the previously cited examples of the IoT seem to already assume the network as a point of contention.  This is likely why constructs such as the Hue gateway, processing onboard 787s, and distributed data management in close proximity to telescopes exist.  Essentially, there appears to be a realization, with current IoT activities, that all data cannot be stuffed into and processed by Internet accessible remotely hosted centralized Cloud Computing platforms.  If that is the case, the author would like to return to the fundamental premise of the paper: does the Internet of Things imply an end to Cloud Computing?  If we look again at the chosen Cloud Computing definition we find that implementation of a platform doesn’t dictate Internet accessibility.  Instead the definition states that, “…on-demand network access to a shared pool of configurable computing resources” is required [4].  This suggests implementations of cloud could occur almost anywhere at any scale including inside an airplane or near radio telescopes.  However, it is likely that Carr’s hypothesis cannot be assumed true.  That is because Carr implies Cloud Computing will shift Information Technology out of local data centers into centralized mega cloud data centers just like a purely centralized power utility.  Given the power industry’s shift towards decentralized smart grids and early signs of the IoT adopting gateway patterns, Carr’s hypothesis needs adjustment.  To clarify the shifts that will happen to Cloud Computing, when IoT becomes far more prevalent, the NIST definition must be modified.  In that spirit, the I'd like to propose the following revised definition – notice that strike through and bold red colored words are used to intentionally expose the difference to the NIST definition.

Cloud Computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing (e.g., networks, servers, storage, applications, and services) and data resources (e.g. sensor data, images, audio data, etc.).  Where computing resources can be managed rapidly provisioned and released with minimal management effort or service provider interaction, and data sources can quickly acquired with minimal network impact. This cloud model is composed of five essential characteristics, three service models, and four deployment models [4].

 

 


The relatively minor changes to the definition should account for a future where the IoT plays a major role.  Yet the I have some humble advice to give to Carr: maybe it is time for a second edition.

 


Image & Table Credits:

  1. A view of the smart grid from the US Department of Energy [33].

Resources

  1. Will Cloud Computing Survive the Internet of Things? (Introduction) - includes all references represented in this post
  2. Will Cloud Computing Survive the Internet of Things? (The Force of Cloud Computing ) - second installment in this series
  3. Will Cloud Computing Survive the Internet of Things? (The Force of Internet of Things) - third installment in this series

Outcomes