Edge Computing - Brief Overview

What is edge computing?

Wikipedia defines edge computing as “distributed computing paradigm which brings computation and data storage closer to the location where it is needed, to improve response times and save bandwidth”, while Gartner refers to it as “a part of a distributed computing topology in which information processing is located close to the edge – where things and people produce or consume that information”, but what does that really mean?

At a basic level, edge computing refers to performing computation (and the data storage required for it) on the device where the data is being gathered rather than in a remote location. Performing the computation process on the device itself (i.e. the edge of our system) allows the device to not only collect data, but also process it locally, without the need to send it to a remote computing center.

In recent years edge computing is developing rapidly, mainly due to the exponential growth of IoT devices, which connect to the internet for either receiving information from the cloud or delivering data back to the cloud. Analyzing the available bandwidth, we’ll find that performing all computing tasks in the cloud will result in quality and latency degradation, but also the costs of the bandwidth required to support that.

Edge Computing vs. Fog Computing

Edge Computing and Fog Computing both deal with processing and screening data prior to its arrival at a data center or Cloud. Technically, Edge Computing is a subdivision of Fog Computing. The primary difference is where the processing takes place.

With Fog Computing, the processing typically happens near the local area network (but technically, can happen anywhere between the edge and a data center/Cloud), using a fog node or an IoT gateway to screen and process data. Edge Computing processes data within the same device, or a nearby one, and uses the communication capabilities of edge gateways or appliances to send the data. (A gateway is a device/node that opens and closes to send and receive data. A gateway node can be part of a network’s “edge.”)

Why do we need edge computing?

There are several reasons for edge computing. The main factors to consider are:

1.      Latency

The ability to process and store data faster, enabling for more efficient real-time applications. Before edge computing, a camera scanning a person’s face for facial recognition would need to run the facial recognition algorithm through a cloud-based service, which would take a lot of time. With an edge computing model, the algorithm could run locally, allowing for almost immediate result. Applications such as healthcare and medical, self-driving cars, smart cities and even building-automation systems require fast processing and response, and will gain great value from latency reduction.

2.      Cost

Cost saving alone can be a driver towards deploying an edge-computing architecture. According to a report by Business Insider, companies waste some $62 billion worldwide on the cloud by paying for capacity they don't need.

Transitioning to the edge reduces the data volume sent to the cloud. Transmitting data from an endpoint to the cloud doesn't come free; the costs include bandwidth, distance traveled, associated network hardware, man-hours to configure and monitor, never mind security data in transit. Less volume translates to less traffic, which in turn reduces overall system cost.

3.      Data security and privacy

Arguably more interesting than financial advantages, edge computing alone allows shifting deeply ingrained business practices. For example, take the standard practice of sending personal data to the cloud, which most organizations do to extract personalized insights via computationally intensive analysis and machine learning. Edge computing allows you to filter sensitive data at the source rather than send it to the data center. Less transfer of sensitive information between devices and the cloud means better security for you and your customers. 

Consider the win-win when edge-level intelligence manages the ability to extract insight for personalization and avoids the vulnerability a centralized cloud repository. Consumers' personal data remains more secure without sacrificing functionality, and organizations can continue to deliver personalization and reduce associated risks, latency and costs. This same shift applies in other areas too, such as sharing compute with external entities, contributing to data marketplaces or configuring compliance into device performance. 

How to implement edge computing in your system?

There is no single recipe for edge computing that fits every company, but a good rule of thumb is that you centralize your computing where you can and distribute it where you must.

Edge computing implementation highly depends on your system architecture, user needs and other constrains. For example, a sensor sampling soil contents and other environmental conditions for an agriculture IoT system, will probably transmit the data to the cloud, in order to reduce HW cost and due to the fact that real-time action is not always necessary. On the other hand, a life-supporting medical device, collecting physiological data from a patient and adjusting drugs and medication dosage, will probably perform most (or all) of the critical calculations by its own, sending the results to the cloud for further long-term analysis, remote access and more.

Summary

I believe that we’ll see a significant growth of edge-computing systems in the near future. It is an excellent solution for businesses needing to analyze data quickly, allowing analyzing mission critical data in real-time and screen data sent to cloud.

In order to design optimal system architecture, you should carefully consider the pros and cons of edge computing for you system and decide where and how to implement it.