IoT Computation Platforms: From Cloud to the Edge
The technology, especially IT and Communication has evolved to an extent that can turn our environment into a fully connected world.
Exabytes of data are being generated by millions of devices giving us the insight that we never had. Such insight can significantly impact the way we live, work, move around or manufacture our goods. Nonetheless, achieving this level of impact is bound to address the problem of how we scale our platform to process such big data.
The workflow of today’s complex data processing applications starts from sending collected data to cloud servers in order to use the cloud data processing services (i.e. computation power) as well as many off the shelve and ready to use applications. Cloud computing can bring benefits to enterprises and IoT platforms by reducing IT infrastructure challenges. However, there are also limitations associated with it. This architecture, by nature, introduces high latency, high costs, and data privacy issues. These are amongst the main factors serving as a bottleneck to unleash the full potential of a connected world.
Edge/Fog Computing: the reverse workflow
Edge/Fog Computing is a paradigm that brings cloud capabilities (i.e. storage, computation, and analytics) to closer proximity where data is collected or the insight is needed. With Edge and Fog computing, one can picture a future that data is kept where it belongs (or at least as close as possible), pull and deploy the application next to data and execute the processes with full control and transparency.
This phenomenon seems unavoidable with the advancement in IoT. It is forecasted that an ever-increasing amount of computation tasks will take place at the edge of the network. Edge and Fog computing by nature will address a considerable amount of latency, bandwidth and cost issues. However, there still remain challenges that are yet not addressed. More on the story behind Edge and Fog and the current challenges are covered in this article.
For executing and scaling applications on the edge, simply adding more processing nodes to our network or connecting the existing ones would not be enough. Dynamic and optimized ways of orchestration of data and processes are key components to this network. On the other hand, different IoT use cases require different applications. The ability to develop such applications in an easy and convenient manner is another important factor. Such applications make use of the data generated by the IoT devices and turn them into smart actions.
Worth mentioning, many Edge computing solutions that are currently available are not mature enough to be used on a large scale. Plus, looking closely at those solutions, issues such as network fragility, maintenance cost, security, and privacy are yet to be addressed.
Furthermore, it is crucial to understand that on a large scale and multi-party scenarios, each edge device (node) has to be somehow assured and incentivized to join such a network. In other word to offer the computation power, storage, and potentially the data, full control, and transparency for the owner is highly needed.
Many of the above-mentioned points are typical issues of any (untrusted) peer to peer network. For the past years, computer science communities are activly trying to step in and address these challenges. In a future post, I will try to paint a picture that describes the essence of a potential comprehensive solution.
DISCLAIMER: This post just reflects the author's personal opinion, not any other organization. This is not official advice. The author is not responsible for any decisions that readers choose to make.
