by Dr. Mung Chiang, John A. Edwardson Dean of the College of Engineering and the Roscoe H. George Professor of Electrical and Computer Engineering, Purdue University
This blog post originally appeared on VMblog
As the IoT expands into every corner of the world, there is no question that intelligent edge architecture is critical to ensure the viability of the enterprise. And not just critical, but unavoidable.
While traditional enterprises have embraced the cloud computing paradigm, significant computing resources still need to be deployed locally to handle tasks that are not suitable for the cloud, or augment the work of the cloud.
A distinguishing characteristic of the IoT is that there will be billions of devices generating exabytes of data, but they will have little computational ability of their own.
There is no feasible architecture that could get that data to a central data center, process it, and get the pertinent results back to the devices in any reasonable amount of time. The network bandwidth requirements would be prohibitive, assuming the capacity existed at all.
Why Edge Has Momentum
But the world of industrial control systems demands latency of real-time or near real-time, certainly no more than a few milliseconds. Consider drone flight control, autonomous vehicle-to-vehicle communications, or smart city emergency systems. These requirements fall well outside what traditional cloud architectures can bring to the table.
At the core of this very real challenge are the decisions about where to compute and where to store data on a substrate of variably available nodes. We can view this as a service continuum between the cloud and the things.
To understand the edge deployment hierarchy, take the simple example of a cell phone providing the intelligent edge for a user’s wearable devices, such as a Fit Bit. When that user is in her car, the car can become the edge for the phone, allowing many functions to be moved to the vehicle. Roadside traffic control equipment can in turn provide edge services to the car, providing both computational resources and additional functionality.
The Future of Edge
Intelligent edge computing needs to be viewed as a complement to the cloud, and can be characterized along three main dimensions:
- Carrying out a substantial amount of data storage at or near the end user.
- Carrying out a substantial amount of computing and control functions at or near the end user.
- Carrying out a substantial amount of communications and networking at or near the end user.
Numerous challenges lie ahead – not only the network bandwidth constraints, latency requirements and resource-constrained devices, but the numerous security challenges and unique user requirements due to the fact that much of the IoT will be comprised of physical systems.
Over time, cloud and intelligent edge could very well converge into unified end-to-end platforms offering integrated services that combine resources in the cloud, the intelligent edge and the IoT.
Dr. Chiang is the John A. Edwardson Dean of the College of Engineering at Purdue University. He is the Founding Director of the Princeton Edge Lab. A founder of the OpenFog Consortium, he now serves on the Steering Committee of the Industrial Internet Consortium, now incorporating OpenFog.