Edge computing is in vogue. In fact, it’s so much in vogue that it can be easy to believe that edge is making “older” computing paradigms, like the cloud, obsolete. There are plenty of comparisons out there that depict edge as “an alternative approach to the cloud” or suggest that edge computing will replace cloud computing. In reality, the idea that edge computing and cloud computing are diametrically opposed misses the point of edge computing. You shouldn’t think of edge as an alternative to or replacement for traditional cloud. It’s not “edge computing vs. cloud computing.” Think of edge instead as an extension of cloud architectures that solves many of the core limitations of conventional cloud computing.
What Is Edge Computing?
As you likely know by now, thanks to all of the hype surrounding edge, edge computing is a type of architecture in which data and processing power are hosted as close to end users as possible.
Edge architectures can be implemented in many ways. You can build a cloud edge, which means workloads are hosted in geographically data centers that bring applications and data closer to users than conventional, centralized clouds. You could also create a device edge that offloads processing and data hosting directly to end-user edge devices, such as IoT equipment.
Edge Computing vs. Cloud Computing
At a high level, it’s easy to make a distinction between edge computing and cloud computing. If the cloud centralizes data storage and processing power, edge computing decentralizes those resources by distributing them to the “edge” of the network, meaning to locations that are closer to end users.
But when you dive deeper into edge and cloud architectures, the lines separating them can become blurry. For example, if you build a hybrid cloud where some of your workloads run in the public cloud but others are hosted in local data centers for security or privacy reasons, does that count as an edge architecture? Arguably, it does, in the sense that some data and processing happens in data centers that are closer to your users than centralized locations in the cloud. On the other hand, bringing workloads closer to users is not usually the primary focus of hybrid cloud, so few people usually think of hybrid and edge as one and the same.
You can ask similar questions about a cloud architecture in which you distribute redundant instances of workloads across multiple cloud data regions to increase availability. Such a setup would have the effect of placing workloads closer to some end users than they would otherwise be if everything was hosted in a single cloud region whose data centers were distant from most of your users. But here again, few would consider this an edge architecture; they would call it a multi-region public cloud.
Using a content delivery network (CDN) places you in a similar boat. If you distribute cloud workloads across a CDN network, is that an edge architecture, or is it just a cloud that takes advantage of a CDN?
Edge as an Extension of Cloud
Faced with quandaries like these, you could draw two conclusions about the edge computing vs. cloud computing issue.
One is that edge computing is just an over-hyped buzzword to refer to practices (like CDN networks and client-side code execution) that are not at all new. I think there is a lot of truth to this viewpoint.
However, I think you can also make the case that if there is anything novel about edge computing as it is being practiced today, it’s the way in which edge is being used to solve the pain points of cloud architectures. This doesn’t mean edge is a replacement for the cloud; on the contrary, edge is a practice that makes the cloud stronger, and allows organizations to use the cloud in situations where a cloud architecture otherwise would not be ideal.
The Achilles’ heel of the cloud has always been the network. The ability to centralize workloads on remote servers, which is what the cloud does so well, has been counterbalanced by the need to connect and access those workloads via the network. When the network is slow, unavailable or expensive, the cloud ceases to work well.
Edge architectures address this challenge by allowing cloud data centers to integrate with edge infrastructure that delivers lower network latency than the conventional cloud. Edge infrastructure may also be more resistant to network failures. For example, an edge constructed of IoT devices may continue to function if its network connection becomes intermittent. And by reducing the amount of data that needs to move between centralized cloud data centers and end user devices, edge infrastructure can also reduce networking costs.
When you pair cloud infrastructure with edge infrastructure, then, you get the agility of the cloud without the drawbacks that typically come with a network-dependent cloud architecture. This is why edge extends and reinforces the power of the cloud, rather than supplanting it.
Edge without Cloud
It’s worth noting that edge and cloud don’t have to go hand-in-hand. You can certainly build an edge infrastructure that doesn’t connect to the cloud in any way.
But this is not generally what businesses are doing. I have yet to hear of a company that ran everything solely on-premises, then decided to migrate to the edge without using a public cloud in any way. Virtually everyone implementing edge strategies today is already in the cloud, and is looking to get more out of the cloud.
In short, edge is a solution that helps organizations double-down on their investment in cloud architectures, not replace them. To talk of “edge computing vs. cloud computing” is to create a false dichotomy that obscures the true value of the edge as it is being implemented today.