I hear it every day now: “We’re moving beyond cloud computing to edge computing.” Pretty hypey, and not at all logical.
Edge computing is a handy trick. It’s the ability to place processing and data retention at a system that’s closer to the target system it’s collecting data for as well as to provide autonomous processing.
The architectural advantages are plenty, including not having to transmit all the data to the back-end systems—typical in the cloud—for processing. This reduces latency and can provide better security and reliability as well.
But, and this is a big “but,” edge computing systems don’t stand alone. Indeed, they work with back-end systems to collect master data and provide deeper processing. This is how edge computing and cloud computing provide a single symbiotic solution. They are not, and will never be, mutually exclusive.
Some best practices are emerging around edge computing that allow enterprises to provide better use of both platforms. Here are a few:
- If possible, edge computing should run all or parts of the same code tree as found in the cloud. This means that if you’re updating the edge-based software, you’re updating the cloud-based software at the same time. Thus, you need to use the same operating systems, processors, and so on both in the cloud and at the edge. Moreover, the software must be aware of where it’s running, on the edge platform or in the cloud.
- Security must be systemic to both the edge and centralized cloud platforms. This means that the security system you’re using for your cloud-based systems needs to extend out to the edge if at all possible. It’s much easier dealing with common security services than security operations that are one-offs for the cloud and each of the many edge computing systems. You’ll lose track quickly, and bad things will happen.
- Just as security needs to be systemic, so does management and monitoring. You’ll find that edge computing systems often go down, typically due to hardware or software failures—moreso than on the cloud. This is due to the fact that they are mostly IoT devices that work in environments that are not controlled. A down edge computing device needs to be rebooted, reworked, or replaced quickly, so the monitoring software should provide alerts ASAP and the cloud-based software be told that the device will not be receiving data.
The edge computing hype will drive confusion in the next few years. To avoid that confusion, you need to understand what roles each type of system plays, and you need to understand that very few technologies take over existing technologies.