It should be no surprise that just as the emergence of cloud gave us cloud-native applications, edge is driving a set of edge-native applications.
These applications, however, will not reside solely at the edge. On the contrary, the accompanying convergence of IT (information technology) and OT (operational technology) is driving new architectural patterns that also take advantage of applications residing in the cloud and data center.
We can see leading indicators of this today when we consider where organizations currently deploy application security and delivery technologies. Less than half (41%) of organizations rely on just one location to host these technologies. The average is just over 2 – 2.11 to be exact. The most likely combination today is cloud and on-premises, but one in ten is already taking advantage of all three locations: data center, cloud, and edge.
This is particularly true for architectures built to support the, on average, ten connected things you can find in most consumer homes. Indeed, if we look at popular edge applications, we discover that they are a mix of cloud and edge, node and endpoint.
We make the distinction between an edge node and an edge endpoint based on the unique properties of each. This helps to delineate responsibility for specific functions that make up an "edge-native application" as each location has unique capabilities and constraints. For example, generally an edge endpoint does not support storage of significant volumes of data. Edge nodes might, and the data center/cloud certainly does. These capabilities and constraints help to shape edge-native applications in ways that use each location’s strengths to benefit both provider and consumer.
This is the fundamental reason that cloud and data centers play a significant role in edge-native applications.
Consider, for example, my Plex media server. I have a local server (edge node) in my house, with an app on my smart tv (edge endpoint). The edge node talks to the cloud-based Plex service for billing and account management, remote access, and for software updates. Other conversations and data exchanges take place between the two that describe usage and performance to help improve the overall application.
It would make zero sense for billing and account management functions to reside on the edge node, let alone the edge endpoint. Similarly, it wouldn’t be efficient (or very safe) to give the Plex service access to my local repository of digital content. Rather, that responsibility is left to the local media server running on the edge node.
In many ways, this pattern is no different than those produced by modernization efforts to provide modern interfaces for traditional (legacy, vintage, retro, mature, choose your euphemism for ‘before my time’) applications. Consider screen scraping text-based systems to provide web-based access at the turn of the century. Or the use of APIs to facilitate mobile banking, which relies on traditional transactional systems. Modernization efforts often produce application patterns that assign responsibility to systems and applications across data center, cloud, and edge boundaries.
These kinds of decisions are being made every day for connected things, sensors, and other edge applications. These decisions emerge as common, edge-native application patterns.
Nearly every one of them includes functions that reside in a cloud / data center.
Unlike the rush to declare data centers obsolete when cloud computing hit the scene, it is unlikely you will hear anyone declaring the same about cloud now that edge is emerging. On the contrary, cloud and data centers will continue to play an integral part of edge-native applications.