With AWS poised to become a $13 billion business, cloud computing has overtaken the enterprise computing industry. However, it’s no secret to insiders that the next disruption is already on its way. IoT gateways and routers that can support edge computing are in the works by Dell and Intel, while software companies are beginning to scale and develop products like Apache Spark to be used in edge computing. The rise of edge computing is here, thanks to the increasingly ubiquitous Internet of Things and how it’s changing the game for data processing.
IoT Problems with Cloud Computing
The rise of edge computing has resulted from a shift in the way users interact with edge devices. Whereas we used to merely consume information at the edge, increasingly dynamic apps and websites have shifted interaction toward data consumption and production at the edge. This trend has only exploded with the boom of IoT devices. Cisco predicts there will be 50 billion things connected to the Internet by 2020, and many of these devices rely on data-hungry machine learning techniques to make decisions without user intervention. With a growing number of things, each creating and consuming huge amounts of data, cloud computing is beginning to face some significant challenges:
- Time: The further away computation is from a device, the longer it takes for information to be transferred to that device. As a result, cloud compute is a poor choice for IoTs running time-critical functions, such as self-driving cars, which can’t operate without instant access to information requests.
- Privacy: With mobile applications currently, end users’ data is collected at the edge, but stored and analyzed by service providers. This leaves users without ownership and control of information that is increasingly sensitive, as IoTs begin to arise for use in the medical and healthcare fields.
- Energy: Pushing large amounts of data out to the cloud and back requires massive bandwidth and much more energy than it would take to compute them at the edge, so energy could be conserved by offloading from the cloud. The rise of edge computing can allow for an efficiency that is critically important to energy-constrained IoTs.
Applications of Edge Computing
Just as the cloud allowed for the rise of new markets like hyperconverged storage and software-defined networking, the rise of edge computing will allow for new applications of technology:
- Smart Homes: Smart appliances like virtual assistants and light bulbs are already being rolled out for use in the home, but the implications of edge compute could allow for even more interactive devices. While the current smart devices on the market respond to voice commands, edge-compute smart devices could potentially respond to conditions in the home through the use of sensors and controllers deployed in various areas of the house. And while uploading large quantities of information about what’s happening in one’s home to the cloud would be highly undesirable for the privacy-minded and clog Internet bandwidth, processing this data locally through edge computing would circumvent these problems.
- Smart Cities: Although the size of most metropolitan cities would prohibit smart operations based in the cloud, the rise of smart cities can be enabled by edge computing. Furthermore, time-sensitive matters like health emergencies and traffic control would not have the latency problems that would arise from cloud computing. Cities like Palo Alto, California are already making use of this opportunity with a $3 million smart traffic light project that aims to prevent cars from sitting at empty intersections.
- Collaborative Edge: Because of the privacy concerns inherent in cloud computing, the data owned by various stakeholders in a given industry is rarely shared, but collaboration across fields like health management could benefit greatly from doing so. By allowing for local data computation to remove or censor sensitive data, edge compute could enable collaboration across various stakeholders involved in heathcare such as insurance companies, hospitals, pharmacies, the pharmaceutical industry, and the government.
Challenges for Edge Computing
Although many of the problems IoTs and dynamic applications are beginning to face with the cloud will be solved by the rise of edge computing, there are still a number of issues that need to be addressed in this emerging industry:
- Security: Although sensitive data from health monitors and security cameras would not need to travel as far from the source (thus preventing certain attacks), an absence of IoT regulation, energy-constrained things, and a dynamic environment at the edge are all security challenges that will need to be tackled as edge computing comes into its own.
- Programming: Although cloud programmers are able to write a program to run in the cloud using one programming language for a single target platform, edge programmers will face the challenge of writing applications for edge nodes with variety of platforms and different runtimes.
- Naming: As IoTs continue to proliferate, an efficient naming mechanism will need to be determined for edge compute for programming, communication, and identification. This system will need to be scalable to suit the large and growing number of IoTs, flexible enough to handle IoTs’ mobility, and highly secure and sensitive to private data. In addition, this naming system will need to be user-friendly for anyone wishing to monitor or control a variety of devices, such as a smart-home owner.
- Service Management: Edge computing will need to not only prioritize certain functions for various devices (data logging for a heart monitor, for example, would not be nearly so important as heart detection failure), but will need to isolate each device so that in a given environment, such as a smart home, a single device crash will not lead to a failure of every device that shares its data resources.
Changes to the Industry
To some, this could seem like the cloud computing industry is poised to implode almost as soon as it’s come into its own. Due to AWS’s success with cloud computing, many companies are still investing massive resources in it, with Oracle, Microsoft, and Google among them. Will the rise of edge computing make these investments moot?
Not necessarily. Although the rise of edge computing will shift many computing tasks away from the cloud, it’s unlikely to altogether replace it. The edge computing paradigm isn’t so much built around eliminating the need for the cloud as streamlining the amount of information sent to and from it to maximize efficiency. For example, while using video analytics to search for a missing child would be impossible through the cloud due to privacy concerns, edge computing could perform the task by searching local data on devices, then deliver the results back to the cloud. In this case, the cloud would perform the task of gathering and processing data sets that have been processed at various local sources, rather than doing the time-consuming work of analyzing each individual video itself. In addition, historical data or large data sets within the cloud can be used to tweak functionality of edge computing processes or leveraged to improve the reliability of data sensing and communication.
A study by IDC suggests that by 2020, ten percent of the world’s data will be produced by edge devices. While the cloud may not be going away, it’s certain that computing will need to adapt to this shift in where data is being produced in order to save energy, decrease latency, and ensure privacy.
