The Edge Will Complete The Cloud
Author: Derek Mak Date: March 14, 2020
In the Spring of 2017, Tom Bittman, VP of Gartner published a blog titled “The Edge Will Eat The Cloud”. What has happened since his assertion? According to statistics pulled from Google searches, such as one below, the number of connected devices has gone from 28.4B in 2017 to a forecast of 50.1B by 2020, a whopping 76.4% increase. One obvious implication is that data generated by these connected devices will need to be captured and processed at the edge. For the most part, the edge infrastructure installed base today is not equipped with the processing and storage capabilities needed.A recent headline of $145m Series C investment in Pensando Systems is only the beginning of the wave of an accelerating edge computing market expected to reach $6.95B between 2019-2024, a 34% CAGR. Does it mean cloud computing is heading to sunset? In my view, the emergence of edge computing will not be a paradigm shift in computing architecture. It is the next evolution! It is not “The Edge Will Eat the Cloud” but “The Edge Will Complete the Cloud”. Both will coexist and complement each other. The edge computing architecture is a distributed architecture where processing can is done closest to the source of the data. It is a massively scalable architecture that will enable low latency, milliseconds response time, data privacy, and the richness of data at the edge in applications that it is needed. The cloud computing architecture, however, isn’t going away. Cloud computing is a centralized computing architecture. It is a massively elastic architecture for computing workload that requires economies of scale. In the world of IoT, both are needed. I would call the combination a Mesh Computing architecture. There is not going to be one model vs the other. Processing IoT edge data requires a distributed architecture while analyzing IoT edge data is much more efficient to be done in the cloud.
In my previous blog, I advocated the creation of micro-data lakes at the edge. Why? Because when IoT edge data is distributed, collection, normalization, encryption should be done at the edge. But where is the edge? Two years ago, I was talking to a CIO of an oil refinery company about their edge analytics use cases. He posted an interesting, but super obvious question. Where is the “edge” in your definition? Being a network guy, the answer was obvious to me. The edge is at the IoT gateway where all connected devices converge. It was a clarifying question for him to put things in context. In retrospect, his question was not that simple. If I were an IoT device supplier making sensor, for example, the edge would be the IoT device itself. The sensor would sense, collect, and even store data. On the other hand, if I am talking to a telecom service provider, the edge could be the cell tower or a service POP (point-of-presence). If I am talking to a bank, the edge could very well be a branch.
There are multiple interpretations of the edge. Regardless of the frame of reference, the goal remains the same for investing in IoT projects, to understand data coming from connected assets. So, in which cases should you choose the use of edge over the cloud? How will the edge complete the cloud? In the wake of endpoints proliferation and the need for low latency applications utilizing rich data sets only available at the source, a distributed data processing architecture is more appropriate than the centralized cloud model. Let me illustrate my thesis here with an example.
Enter a 21st-century grocery store, where check-out lines have all but disappeared (e.g. Amazon Go). In such stores, there are WiFi access points, security cameras, RFID sensors, temperature sensors, humidity sensors, proximity sensors, and a hose of other connected devices working together to create a frictionless shopping experience. All these connected devices and sensors are generating data continuously. Some of the data get sent to the cloud; some just be overwritten on the devices with new data. The point or device, where data is processed at the store is what I would call the edge. There are many devices like this, such as device gateways. In order to create a truly frictionless shopping experience, a system (hardware + software + services) must be in place to ‘sense and react” to events occurring in real-time at the store. For example, someone walks into the store to purchase a carton of milk, a dozen eggs, and a package of bacon. It is great that there are no longer check-out lines, the store has eliminated the wait time which is a huge source of frustration for our impatient generation. However, the #1 source of frustration is my experience is actually locating the items you need at the right aisle, on the right shelf, with the least amount of time. That is a frictionless moment the digital grocery store must enable.
With edge computing, it is now possible to bring that frictionless moment in real-time. In addition, the digital grocery store has all the data at the edge and knows that the person is picking up a carton of milk, a dozen eggs, and a package of bacon. He might enjoy a fresh loaf of French bread and other items that the grocery might want to suggest and incentisize with instant discount coupons. The store can direct him to get the suggested items through the augmented reality capability on his mobile phone without adding many delays to the shopping experience. Well, this is not something you can do without analytic modeling in the cloud. The product purchase affinity analytics, once developed in the cloud, can be pushed to the edge to provide the “sense and react” function. This is just a simple example of how “The Edge will Complete the Cloud”.
The “sense and react” capability requires rich data collection and real-time processing at the edge. So far, my example only shows the consumer experience. There is a supply side of the equation, which is equally important to address. To show how The Edge will Complete the Cloud for the supply side of the equation, let me expand the digital grocery store example to a retail chain, say Walmart, with thousands of suppliers who want access to their product sales data so they too can perform the analytics needed to up their game on their digital customer experience. Walmart has been providing detailed sales data to its suppliers for decades. The difference now is the richness and timeliness of data. While Walmart can continue to provide suppliers access to sales performance data through the cloud in a centralized architecture, the ability to access edge data opens new possibilities for both the retailer and the suppliers – it gives the retailer a new avenue to monetizing its data while providing suppliers the ability to target consumers in real-time within the boundary of personal data privacy.
To make this type of multi-party digital interaction seamless, you must harness the end-to-end data management from the Edge to the Cloud and vice versa. It requires a mechanism (and incentives) to securely exchange data amongst all stakeholders. I brought up a few examples of data exchange platforms in my last blog. Without over-analyzing each of their capabilities, their approaches are all premise on storing the data in the cloud, their cloud. As I have illustrated, both Edge and Cloud must co-exist to deliver the frictionless digital experience, not only for the retail example I shared. It applies to all industries undergoing digital transformation. Every industry, every business has its “edge”. In manufacturing, their edge is at the factories. In oil & gas, their edge is at the refineries and the oil extraction sites. In public transit, their edge is at the stations, the trains, and the buses. In banking, their edge is at the branches. In utilities, their edge is at the substations. In airlines, their edge is at the airports and the planes. Where is your edge? I have spent a fair amount of time thinking about our future, our planet, and our environment. Because what good is to possess IoT, edge computing, cloud, artificial intelligence, and all the other technologies if they are not used to positively alter the course of damage we inflected onto our environment? An estimate of eight million tons of plastic waste enters the ocean every year. That is equivalent to disposing four million vehicles into the ocean every year! Can we create a better world for the next generation? I believe there is great potential here. Stay tuned for my next post – Bridge to a Sustainable Future.