Let me start this with some of the most common questions asked these days; ‘Is cloud going away?’ ‘What will happen after the sunset of cloud charisma?’ Cloud computing is popular now, and it’s imperative for people to ask such questions. The simplest answer is cloud computing is not going anywhere. On the other hand, it will pave way for others as a big daddy so that we can embrace digital and IoT better. How can this be done? One of the ways is Edge Computing which is expected to act complementary or in continuation to cloud.
Edge Computing is all about having processing and analytics closer to the data source or the endpoint for enabling real-time or near-real time insights. It is the reason why IoT application is a prefect use-case for it. For instance, a smart engine or a locomotive would require real-time insights for its movement (speed, direction, and correct way acting as other variables). The requirement to have real-time insights considering multiple variables would require data sets to be processed, analysed, and acted upon swiftly, without any latencies. Latencies might depend on thousands of data generating chips with varying volume, velocity, and variety that would make it even more difficult.
The entire loop of collecting, analysing, and feeding back of data to these devices for their smooth functioning and without any glitch requires zero latency compared to the cloud-based analysis which is being prescribed today as part of cloud computing solution. Similarly, an AI-based drone/robot will need to perform like a human being without wasting time on analysis. This paves the need for a new architecture. Will this architecture have room for current version of cloud computing? Most certainly.
All the data that will be processed at the edge or at the entity itself (acting as an edge) would need further analysis in terms of its health, pattern, trends, and improvement aspects. This can be done with the current cloud formats as it can withstand latencies without any direct effect on the immediate functioning of these entities.
The second aspect would be with the storage of such huge volumes of data wherein not all data would be relevant for the storage. However, a tiny fraction of that data multiplied with the data sources would certainly be more than what can be handled now. Cloud, in its current form, would again be the lone bearer in handling such a demand.
How soon can we expect the Edge? Well, it’s the maturity of use cases such as IoT, collaborative applications, AI based scenarios or simply the ecosystem gearing up to leverage such architecture in its entirety.
Another area where Edge architectures would do wonders is that of intuitive/interactive customer experience, context aware services, and associated applications that would embrace Edge-based architectures to have ‘lightening’ decision making capabilities. These certainly cannot be served with current service models owing to latency, efficiency, and cost issues. Considerable performance upgrades and efficiency is expected as data/insights move closer to the endpoint.
Now the question arises, will cost be a driver when adopting edge computing? Yes, it will be a driver as there is a significant chunk of data that moves to-and-fro between the source and the cloud. It would certainly be reduced, either in terms of volume or in terms of number of to-and-fro counts or both, thereby initiating a saving on bandwidth and storage capacities, not to mention the savings in time/effort in trying to do it all with the current models.
Hence, what does it take to be on the Edge? Is it technology alone? This would require as much as an architectural change as process, people, and services transformation. For instance, in terms of processes, it would require the segregation of functions/insights that are required for smooth operations versus the associated processes of conceptualization, creation, delivery, and service that needs change.
Similarly, the support services, skills, performance metrics, hierarchies, structures would need a change to support the ecosystem. Also, skills requirement would drastically change from traditional SME based model to more cross-skilled and automation based setups. Human intervention would be required but would be limited. And similarly, would demand denser skills than the L1, L2 skills that are more prevalent today in the industry.
So, what will this Edge be? The architecture needs to be modified or in some cases, would change completely to support it. Edge itself may act as another cloud where, in some cases, local processing will happen either on a semi-mobile or fixed device that will be feeding into the Edge directly. This could simply be the entity or device itself owing to technological advancements and requirements or a flurry of micro-datacentres acting as Edge gateways.
Their placement or identification will depend on what kind of application and performance is intended, and also the location and volume of customers sometimes. In some cases, it may simply percolate down to the usual run-of-the-mill applications as well to manage latency issues. This could also involve interactive experience requirements and connected devices generating massive amounts of data that need both warm and cold insights.
Lastly, all this connectivity may draw flank on security issues. As security vulnerabilities and data privacy issues may arise due to more ‘intelligent’ architectures being more prone to exploitation or acting as extended attack vectors. For edge architectures to prevail, security and privacy setups may also need to become smarter in order to address the ‘intelligent’ questions posed by such implementations.
This would need new administration and governance models along with up-skilling of resources towards management and analysis rather than maintenance alone. To be on the Edge, one would need support from these pillars so as to not end-up being on the wrong side of it.