Aside from the lure of the public clouds, which got bigger and stronger in 2020 amid the COVID-19 scourge, the world of data and analytics became a lot more distributed last year. While it seems clear that most AI models will be trained on the cloud’s behemoth clusters going forward, just about everything else about data and analytics is being conducted in places far and wide.
The edge, as it’s commonly defined, is everywhere in the world that is not your primary data center. In 2021, your primary data center is more likely one of the three big cloud providers, Amazon Web Services, Microsoft Azure, or Google Cloud. According to a recent report by Gartner, spending on public clouds is expected to rise 23% this year, to $332 billion. That’s a healthy chunk of the overall $3.8 trillion that companies are expected to spend on IT and telecommunications in 2021, per Gartner.
Spending on the edge, by comparison, is a bit smaller, but of the same magnitude. A September report by IDC concluded that spending on the edge will hit $251 billion by 2024, representing a compound annual growth rate of 12% from 2019 to 2024. The key drivers of that growth in edge computing include demand among consumers for better experiences, along with a faster cycle of data-driven decision-making, IDC says.
For example, on an oil rig, companies can’t afford to wait for data coming from sensors to be routed back to the cloud or the headquarters for analysis, Gartner Donald Fienberg said his “Top Trends in Data and Analytics, 2021” session at the recent Data & Analytics Summit.
“I don’t want to have to send it back to the home office to get the result and then send back an alert that says, uh oh you’ve got a real problem,” Feinberg says. “By then, the oil rig already blew up. It’s that simple.”
Yes, But Which Edge?
The location of the edge seems obvious when the data has to move hundreds or thousands of miles to be processed, as is the case on an oil rig, or an offshore windfarm, such as those monitored by GE Digital.
But when the edge comes up in conversation, it’s not always clear which edge the speaker is talking about. According to Forrester Research, there are four distinct edges, including the enterprise edge, the operations edge, the engagement edge, and the provider edge.
Each of these edges has different characteristics and serves different needs, the analyst group says in its recent report, “The Four Edges of Edge Computing.” For example, the enterprise edge could include second- and third-tier data centers, co-location facilities, office spaces, and even smart buildings. These edge locations run the same software as is run in the “core,” but it’s just hosted in a more “distributed data center mesh,” the analyst group says.
The operations edge, by comparison, could include local networks of smart “things” (like watches or time clocks) and remotely deployed gateways in commercial and industrial spaces. These edges tend to be more remote, and may use low-power networks and cellular broadband to connect to the core.
Engagement edges can be thought of as far-flung services offered by vendors, such as a smart home automation solution or employee engagement. Vendors will often deploy dozens or hundreds of clusters around the world to support these engagement edges, which provide predictive capabilities for data originating from IoT devices, such as video cameras in a home security deployment, according to Forrester.
The provider edge primarily will serve emerging telecommunications requirements, such as 5G cellular networks. The provider edge will eventually be capable of hosting operations and engagement edge solutions, such as content delivery services, Forrester says.
Data Fabrics, Etc.
Creation and consumption of data on the edge is making the analytics environment increasingly fragmented, according to Gartner analyst Rita Sallam, who presented with her colleague Feinberg in the “Top Trends” session.
“It’s distributed everything,” she says. “Data is increasingly distributed. People are working remotely. Customers are increasingly online and emancipated. And devices at the edge are growing. And users of analytics at the edge are consumers of data and analytics across our enterprise, all of which represents new opportunities for competitive differentiation and operationalizing business value.”
The growth of edge computing is tied up with other major computing trends, including the rise of composable data and analytics. Not long ago, companies could have expected to conduct their data and analytics processes within a limited number of tools and technologies. They would have used a limited number of techniques to achieve their goals.
That is definitely not the case today, as the whole data and analytics ecosystem has blown up. Being able to piece together a composite solution from a number of components and techniques, most likely linked together with microservices and running in containers–and of course, acting on data that is collected from the edge–is seen as the key to thriving in this new analytics environment.
Underlying the edge and composable data and analytics is another major technology trend: the rise of data fabrics. According to Feinberg, data fabrics abstract away the physical storage of data and help users build applications in a low-code manner.
“Data fabric is simply data from everywhere,” he says. “You should not have to worry about where it is, how to access it, or anything like that. It’s handled for you by a solution or an architecture that we call data fabric. It’s not a single tool, but in fact a set of tools that are put together into a solution.”
At their core, data fabrics are held together with graphs, which can keep track of and serve metadata corresponding to the various entities involved (this poses a challenge to data governance, but the vendors are stepping up, Feinberg said). It’s all part of the next-generation of data operations (DataOps), or Xops, as Gartner likes to call it.
There are a lot of technologies involved with edge computing, and they are evolving at a breakneck pace at the moment. But if companies take the time to piece it all together–and above all, find the folks with the right skills to build it and manage it–then it’s all worth it in the end, Feinberg promises.
“Organizations, vendors are putting together complete solution that will get us there,” he says. “We believe the drivers far outweigh the inhibitors. The inhibitors are diminishing. The drivers are getting better and stronger and therefore it’s becoming more useful.”
Composite AI: What Is It, and Why You Need It
Hitching a Ride to the Edge with Akamai
ASICs on the Edge Help GE Digital Optimize Energy