Just as we’ve become comfortable with cloud computing, here comes edge computing, ushering in a new set of uncertainties and challenges. But, as we are prone to do in the technology world, we just look at it is as another nail, and apply our cloud computing hammer to it. We did it with mainframes and client server and virtualization and cloud computing itself.
And we’re doing it with edge computing. Take IoT for example. The current model for the large cloud players is to “create in the cloud” and “push to the edge.” This works fairly well for traditional IoT devices. Consider the following model.
IoT devices have typically been sensors or data gatherers that collect some data and send it somewhere for processing – such as billing or maybe trend analysis, or some other “backend” function. Not much happened at the edge. Maybe just some data that trickled back to IoT devices to reconfigure them, or maybe set a new threshold. The complexity in this model was really about scaling the backend, central cloud.
Today’s edge computing model is significantly different. Edge/IoT devices are getting really smart, and they are creating lots of data. Massive amounts of data in fact. Some of that data needs to be processed at the edge, but a lot of it sill needs to go back to a central cloud for a new set of backend functions such as deep analytics and machine learning. The results of that backend processing then get pushed back to the edge and used in closed loop processing, as shown in the following diagram.
The complexity has shifted from the central cloud to the edge. The applications are much more sophisticated and utilize tight OODA (Observe, Orient, Decide, Act) loops to perform time-sensitive functions. And, they work together with other applications and devices in a sophisticated ecosystem that now requires “East-West” connectivity, in addition to the traditional “North-South” connectivity to the backend systems. Awareness of the ecosystem, and its complex dependencies and relationships, becomes critically important in the edge computing model.
Here’s the rub though. The current model offered by large cloud players doesn’t comprehend the edge complexity. Dealing with the integration of multiple edge computing elements (applications, devices, networking, virtualization) requires an orchestration model that leverages a central view of the all the components that make up a solution, and then uses domain-specific controllers that are optimized for each of the edge computing domains. Only then can you create a truly integrated edge computing ecosystem.
That’s what we do at CPLANE.ai. Contact us to learn how we can accelerate your edge computing strategy.