Organizations are hearing a great deal about edge computing and how it is the future for expanding their computing architecture.
The boundaries of enterprise infrastructure are on the move, and the data center will simply no longer be the center of data, according to research firm Gartner, Inc. More and more data will stay closer to “things,” and closer to the enterprise itself – this is the nature of edge computing.
Edge computing is a decentralized complement to the cloud for delivering on the promise of digital business and enabling IoT and more immersive experiences. The edge and the cloud work better together, allowing CIOs to expand the reach and capabilities of modern IT. In fact, the edge is expected to grow faster than today’s cloud and become critical to every enterprise.
To better understand how edge computing works and the impact it will have on organizations, Information Management spoke with Gartner analyst Thomas Bittman. He discussed the future of edge computing, and specifically:
- How disruptive technologies will change computing
- Enterprise readiness for the growth of edge computing
- How digital business will push infrastructures to the edge
- The benefits of edge computing and why the focus will shift from cloud to edge
Information Management: Interest in edge computing is rapidly growing. What exactly are we talking about when we refer to “edge” computing?
Thomas Bittman: The edge is the physical location where things and people connect to the networked digital world. Importantly, we distinguish between things (and the IoT trend) and people (and a trend toward more real-time, natural, immersive experiences). Edge computing is processing located close to things and people.
“Close” is a fuzzy word, but edge computing is solving a few problems:
- Speed of light: Some use cases require more real-time response and can’t wait on network transmission that is too slow or unpredictable. Think problem on rapidly moving assembly line, pressure in a pipeline, Tom Cruise walking by a virtual billboard.
- Cost of bandwidth: As things light up (IoT), they have a lot to say, and need to be told what to do. Sensors in cars, cameras, actuators, lights, valves, etc. We predict that for the first time ever, the majority of enterprise data will be both created and processed outside of data centers and cloud by 2022. Why? Because bandwidth costs do not go down as fast as compute costs. It doesn’t matter if you have 5G or 20G. Data has gravity, and compute will move to it. By the way, CDNs today, caching and delivering data closer to people and businesses, is a mature example of ONE kind of edge computing.
- Connections go down: Leveraging the scale and capability of cloud computing is great, but when the connection goes down (for the oil rig, the ship, the military unit, or the home!), business still needs to run. Edge computing can provide limited autonomy.
- Data intimacy: as more things are sending more data, the data becomes more intimate about people, and about a company. Where my kids are. My heartrate. Whether lights are on and in what room. What equipment is running. Already, enterprises are using edge computing due to regulatory restrictions (such as GDPR) on personal information – for example, facial recognition data used in an amusement park, but not sold or used elsewhere.
These four “problems” are based on actual client inquiry patterns – what clients are saying they will do with edge computing. Some use cases require one of them, some require several.
IM: How does edge computing relate to cloud computing?
Bittman: If you ask a cloud provider, the answer is that edge is the edge of the cloud. If you ask any other vendor, the answer is edge computing complements and integrates with cloud.
We think it is both, but more of the latter. We think that the drive toward digital business paints the picture – digital business is about blurring the physical and digital, and coming up with new business models based on that. It’s about leveraging more interactions between people, businesses and things. It’s about smaller windows of opportunities – business moments – that require more agility closer to people, businesses and things.
Cloud computing and the cloud computing business model is about massive back-end scale and agility. Edge computing complements cloud computing with widely diverse technologies and topologies closer to the edge that enable front-end agility and real-time capability. SOME of this can be enabled by cloud providers extending closer to the edge – regions, metro-area data centers, colos. But the closer you get to things and people, the more use-case specific edge computing needs to be (and the less it fits the cloud provider business model).
Cloud providers will address some large portion of edge computing requirements by extending closer, but most of the requirement will be addressed from the edge toward the cloud, and not the other way around.
IM: Do you have a sense of how many organizations are currently invested in edge computing?
Bittman: It’s very early. Someone asked me recently if this was a ballgame, what inning is it? I said national anthem.
There are some edge computing use cases that already exist – CDNs (as I mentioned), and many ROBO deployments. But the bulk will be with new interactions with things and people. The OT (operational technology) to IT conversion has been going on for more than a decade, and is accelerating as organizations want to improve their broad control and automation of plants and facilities.
But all that said, we are less than 1 percent into the edge computing trend. We believe this is accelerating, though. Our edge computing inquiries are doubling every quarter. We believe the majority of large enterprises will have at least one edge computing use case for IoT or immersive experiences by YE21 – and at least six by YE23.
IM: How about the size of the potential market in terms of organizations that are considering it?
Bittman: We are working on that now, at least to get a very fuzzy answer.
Diversity of use cases is the huge challenge. McKinsey, for example, put out a report a while back that identified more than 100 use cases. We’ve built a model of all potential edge computing use cases based on interactions between people, businesses and things, and have twelve wildly different use case categories.
Edge computing is not a market – it is many markets, and many topologies. It will be delivered in telcos, regional datacenters, micro-datacenters around cell towers and in metro areas, edge servers located in plants and facilities, intelligent gateways, intelligent speakers (e.g., Alexa, Google Assistant) becoming edge computing in homes, smartphones, and embedded.
That said, we think the investment will be very large. I wrote a note a few years ago saying the edge will eat the cloud. It was a maverick note, intentionally controversial. But I will stand by that in one sense – edge is what’s new, and what will differentiate companies, and investment will shift to edge computing as the critical competitive edge. Cloud, in the meantime, remains the important back-end – but not leading edge (sorry, had to say that).
So how “big” is edge computing? Until we’ve done more work on sizing all the markets, I’d say it’s about 0.7 cloud – in other words, probably less than cloud in terms of overall spending, but pretty close. And compared to about 0.0 cloud today, that’s a big leap.
IM: How will the adoption of edge computing change computing strategies?
Bittman: There’s been a march toward eliminating datacenters and pushing workloads to cloud computing providers, effectively ignoring location. Edge computing will force a brand new focus on datacenters that aren’t datacenters – processing locations in the thousands, or millions, where location is paramount.
Anything that has customers involved, or employees, or specific locations, or plants will require edge computing at some level. Except for B2B back-end, massive back-end processing, edge computing will be a part of workloads.
By the way, this will even change machine learning plans – the common thinking is ML takes place where the massive pile of training data is, in the cloud. But when the massive data is location specific and at the edge, ML training will take place there, too. The point is, a huge left turn is coming in terms of how we think about our computing strategies – not away from cloud, but expanding to include the edge, with new expertise needed.
IM: What are the top business factors that are driving organizations to adopt or consider edge computing?
Bittman: Becoming a digital business – and doing business that can only be done with real-time, local, reactive, intelligent edge capability. Reducing cost at remote locations (automation). Competing with locations against strong e-commerce players (think immersive experience in a store). Revenue growth by going after new business opportunities that require faster response. Avoiding disasters. Engaging new generation customers that will expect immersive more and more.
If I can react in real-time with a personal and immersive experience to a customer, and you can’t, I win.
IM: What are the greatest benefits that can be had by organizations that adopt edge computing?
Bittman: Same as above. But let me just add – there is a generational and cultural element here that is critical. Before e-commerce, we were perfectly happy to drive to multiple store before making a purchase decision. Now we expect to be able to find and transact a purchase online in minutes.
But even today, how often do we separately Google search for information on something, check out YouTubes, discuss on our social network? And we do all of that through a very unnatural browser interface. If I can immerse myself, get information by voice, augment the physical world with information about it, there is no going back.
We believe the consumerization of IT effect will be very strong here. Immersive experiences don’t always require a headset, but if I can use my smartphone for AR, enjoy a movie in VR, collaborate with schoolmates on a project in a virtual workroom, manipulate objects in a local metaverse – I will want to do that in my workplace, with my stores, with my work associates. There will be pressure to speed up – and that will require edge computing.
IM: What are the top challenges to achieving desired results with edge computing?
Bittman: There are many challenges, but three stand out:
- Managing the edge: Edge computing (and edge devices) combines the necessities of data center management with the scale of mobile computing, without IT staff at the locations, and with wild diversity of deployments. Managing, provisioning, updating, ensuring resilience, hostile locations (weather, etc.), power – all are huge, new management challenges. Edge as a service – at a variety of layers – will be a major thing because of this. So will appliances tuned to different classes of use cases.
- Securing the edge: Similar to management, edge computing brings the security requirements of the data center to the scale of mobile computing (but with diversity). The edge needs to be protected, and the architecture needs to be protected from the edge.
- Data management: We are moving from data in specific, discrete locations to data droplets all over. We need to manage, process, aggregate, archive, destroy, manipulate data effectively and efficiently at all levels of the compute topology, from the edge to the “core”. Some will do this well, some will do this very badly. While we believe there will be more data at the edge – creating opportunity – the data at the edge is different. On average, data at the edge has a shorter half-life of value (events, happening now), and lower value by bit (think video streams of empty parking lots). We need to preprocess and filter that data effectively to get value, and only bring data back to the core when we can mine more value.
This blog is provided for informational purposes only and may require additional research and substantiation by the end user. In addition, the information is provided “as is” without any warranty or condition of any kind, either express or implied. Use of this information is at the end user’s own risk. CenturyLink does not warrant that the information will meet the end user’s requirements or that the implementation or usage of this information will result in the desired outcome of the end user.