Companies are underestimating what it takes to run edge-based devices. One of the problems is the number of places where they exist: small computers inside an industrial robot, the computing that monitors systems within a jet plane, or even sensors that monitor customer movement throughout a retail store.
Edge computing is hyper-heterogeneous, meaning that most edge devices are different, need to be managed differently, secured differently, store data differently, and run different native operating systems. They are becoming the ultimate in complexity.
Although all this is manageable, it’s not without a commitment of time and money on somebody’s part, and therein exists a forthcoming problem. Who owns the edge?
Let’s take the development of cloud computing which began with “shadow clouds,” or unauthorized clouds, started by sales, marketing, and manufacturing departments that worked outside enterprise IT’s compliance and control. These SaaS, PaaS, and IaaS public cloud services were eventually put back under the control of enterprise IT, which was tasked to maintain and manage systems that they did not originally create.
These days we’re looking at the same issue with the thousands of edge computing systems popping up in enterprises. The owners are the departments that sponsored the development and installation of the edge-based computers, both big and small, complex and simple. We’re getting to the same tipping point as we did with shadow clouds where the departments might be looking to transfer ownership to IT again.
This seems like déjà vu, considering that IT got tasked with migrating to public clouds just a few years ago. They may not have the resources or the desire to take over a rather complex set of systems. For some enterprises, a battle is looming.
I can understand the dilemma. Managing a large number of edge devices that are widely distributed and have different approaches to management, governance, security, and more, is, and will be, a huge effort and expense, perhaps more than was estimated.
My advice is to get ahead of this. Deal with the complexities of managing the devices during the design phase of the edge systems, not as an afterthought. However, I don’t think that most enterprises will think that through—again.