Living on the Edge — All You Need to Know About Edge Computing

IoT has grown at a rapid pace with tremendous innovations cropping up from all across the globe. Many organizations are already harnessing the power of IoT to deliver higher performance, automate processes, enhance efficiency, and reduce cost.

Around 50% of enterprise organizations are planning to implement IoT within the next three years, according to a State of the Network research in 2018. With the increasing number of data devices, the amount of data collected will also increase significantly, which will require enhanced data management and storage platforms. This is where edge computing comes into the picture.

It has become one of the most trending technologies in IT infrastructure and networking. As more connected devices are coming online and placing higher demands on networks, administrators and other IT personnel are struggling to keep their infrastructure perform optimally.

Understanding Edge Computing

Edge computing technology enables easier and faster processing of information to be performed closer to its source (i.e. the location where it is collected), and in turn, this helps companies in analyzing important data in near real-time. Edge computing has appeared as a prospective choice, putting IT resources and apps at the edge of the network, away from the traditional centralized core.

In companies where a massive amount of data is processed, edge computing empowers organizations to quickly process a large number of datasets and information without having spent much resources or efforts. The ‘Edge’ depicts a way to make applications and systems more efficient by shifting data, services, or components from a centralized core and putting them in the close vicinity of the logical extreme, i.e. the edge.

In simple words, the data to be processed is placed closer to its original destination, which reduces the network latency for an overall round trip. Usually, such an architecture requires computing power, storage, and data microservices to be redistributed accordingly. However, it is important to understand that while the performance of certain elements may improve by deploying them near to the edge, others should be placed at the core of the architecture.

Need for Edge Computing

Edge computing has ample of benefits. Supposedly, there is a self-driving car where the car is continuously sending a live stream to its central servers. Now imagine if another vehicle that is moving right ahead of this self-driving car pushes the brakes suddenly, the car will have to take a crucial decision then.

It can’t afford to wait for the central servers to process the information/data and need a quick response. Although there are legit algorithms like YOLO v2 that have increased the speed of the process for object detection, the latency usually occurs when the car has to send terabytes of data to the central server where it is processed, and then receive the response and then take action based on it!

Therefore, it is imperative to have a basic processing system which enables the car to make decisions like when to decelerate or stop by itself in specific situations.

The main aim of Edge Computing is to reduce the latency by incorporating the capabilities of the public cloud to the edge. It can be achieved in two forms — extend the public cloud seamlessly to multiple point-of-presence (PoP) locations and custom software stack emulating the cloud services operating on existing hardware.

Here are some of the major benefits of using Edge Computing:

1. Privacy: Avoid sending all raw information to cloud servers for storage and processing.

2. Reliability: Edge computing is capable of working even disconnected to cloud servers.

3. Real-time responsiveness: Edge computing enables near real-time responsiveness, which comes in handy in several situations.

Let’s take another example to simplify edge computing and understand how it works.

Consider a device which responds to a particular keyword. For example, Jarvis from Iron Man. Imagine if you had your personal Jarvis who sent all your data to a remote server for analysis. But what if it was intelligent enough to respond when it is asked for, quickly. Not only the later option is reliable but also real-time.

Brain Krzanich, CEO of Intel said that self-driving car would generate about 40 terabytes of data for every 8 hours of driving. Now with the data skyrocketing, the time taken for transmission will also lengthen. In cases of autonomous cars, it is highly important to incorporate reliable, real-time, and quick decision-making systems.

Edge computing enables autonomous cars to make quick decisions in the blink of an eye, whether to stop or continue driving. Without having the ability to make proper decisions, technologies like self-driving cars with the foundational unit of artificial intelligence and machine learning can be quite dangerous.

Another common example for Edge computing is drones or quadcopters. Let’s suppose you are a travel blogger who loves traveling to scenic places, from high above the mountain cliffs, to deep in the lush green forests. You make videos and capture photographs using your drone, which helps you get amazing shots.

Though you can fully control the direction and speed of your drone, there are high chances that your drone could crash into obstacles like trees, hills, or buildings. In order to avoid drone accidents, these devices are well-equipped with Edge computing, which enables the drone to identify obstacles in its vicinity and steer clear of the area if extremely close.

Fog Computing and How is it related to Edge Computing?

With Edge Computing creating a buzz all over the business world, another term has been catching on — Fog computing. The fog layer is sandwiched between the edge layer and the cloud layer. It bridges the gap between the two layers.

Edge Computing: It promotes the processing power, communication capabilities, and intelligence of an edge gateway or appliance directly into devices such as programmable automation controllers (PACs).

Fog Computing: It processes data in a fog node or IoT gateway by pushing the intelligence down to the local area network level of network architecture.

Types of Edge Computing

Cloud: Cloud essentially refers to massive data centers run by cloud providers such as Azure, GCP, and AWS but may also include VMware Cloud on AWS and other cloud or hosting providers. The principal features of the cloud are that it can be operated at scale and is a centralized technology.

With cloud type edge computing, you will have access to ample of services and resources along with a substantially high infrastructure. The only drawback associated with Cloud in Edge computing is that it is centralized; latency is higher as network connectivity to devices or sensors is not promised.

Device Edge: Also referred to as nano DC. As implied by its name, nano DC or consists of one or a few small servers that have minimal compute capacity. Servers in these data centers usually cannot be rack-mounted and can run without cooling.

Device edges are found in locations that are not linked with a data center, for example, wind turbines, cars, or factories and can be rugged to handle extreme conditions. One of the major benefits of using device edge computing is that it can be located right next to IoT sensors, and the bandwidth, latency, and connectivity issues are minimal. However, its drawback is what its benefit is — such small devices can provide minimal capacity and services.

Compute Edge: Also known as a micro DC. It is a small data center with the capacity of a few to several racks of servers. Computer edge is usually located near the IoT devices, and it might also be required for local compliance reasons.

The key benefit of computer edge is that these data centers include cooling facilities and consist of standard rack-mounted servers. Though you can easily store many resources on computer edge, the range varies from that of the cloud system.

Sensor: IoT sensors are devices which can either gather data or control a device such as counter, light bulb, security cameras etc. Sensors do not contain any compute capacity. Instead, they communicate with the compute edge, cloud edge or device edge, depending on the latency, connectivity requirements, and bandwidth.

Some Practical Examples: Edges Around You

Edge computing is much more common than you think; here are some examples of edge computing near you:

1. Smart Homes

2. Smart Street Lights

3. Automated Industrial Machines

4. Automated Vehicles (cars, drones etc)

Some of the prominent use cases of Edge computing are Virtual Reality (VR) or Augmented Reality (AR) and Internet of Things (IoT). For example, the excitement and rush that people get while playing an Augmented Reality based game such as Pokemon Go wouldn’t have been the same if “real-timeliness” was not present in the game.

This game was made with edge computing where the real-time events were taking place in the smartphone itself without the need of contacting any central servers. Edge computing also supports machine learning greatly. All the bulky training of ML algorithms can be deployed on edge for near real-time data predictions that can enhance the user experience.

Conclusion

Latency can be reduced significantly by bringing computing closer to the origin of the data. It not only saves time but also enhances the end-user experience, which is a key driver for the success of a business in today’s competitive world.

It’s quite evident that edge computing is becoming an integral part of today’s data-driven technology. The next-gen computing will be heavily influenced by Edge computing. It will continue to expand in different verticals with more innovations driven by precision-based real-time data.

Read More Blogs:

Cloud vs Fog vs Edge Computing: 3 Differences that Matter

Understanding Software Architecture Frameworks — Microservices, Monoliths, SOA, and APIs