Will Edge Computing Render the Cloud Obsolete?

The Edge is making a comeback. No, not the U2 guitarist. We’re talking about the network edge. However, like the bass player, the network edge is about finding ways to delight the enduser. The edge first enjoyed great attention and popularity in the hey day of content delivery networking. The goal of CDN software and services was and still is to cache content at points close to the consumer, to ensure rapid delivery of said content. Back in 1999/2000 CDNs were ensuring the delivery of web pages and short, low-quality videos. The use case discussed, by the myriad of CDN software vendors, was the ability to serve the online Victoria Secret’s fashion show without a glitch – meaning without buffering or pixilation.

We’ve come a long way since then and the network edge is more important than ever. Content Delivery Network providers, like Akamai and Limelight, expect to see an uptick in demand thanks to the volume of video traffic and the Internet of Things. Cisco predicts that 80% of data on broadband and mobile networks will be video, by 2020. A recent study from 451 Research reflects that just under half of the study respondents indicated a preference for performing IoT data processing (e.g., data analysis, data aggregation, data filtering) at the edge. IDC agrees, anticipating that 45% of IoT data will be stored, processed, analyzed and acted upon by edge computing by 2020.

So, why is the edge such a hot commodity when all the chatter has been about the Cloud for the past several years? It’s all about time, money and the promise of personalization.

Let’s consider time. We are an impatient society. The advances in technology that allow us to instantly search and consume all types of content have spoiled us. We stream video content wherever we are. Whether it is news, late night comedy or favorite series or user generated content, we expect to access and consume content on demand. The inability to enjoy content due to technology challenges related to device, network performance, service architecture, is not acceptable.

Behind the scenes, that content is being processed and prepped for delivery to our chosen device. Decisions are made about where to store and transcode that content before distribution. The consumer doesn’t know, nor do they care. They do care that is arrives in a timely and high quality manner.

In the traditional CDN model, popular content is pre-positioned at the network edge. However, increasing consumer demand for streaming video presents persistent challenges in the form of storage, compute and bandwidth costs. The current debate focuses on balancing the need to deliver content in the right format and where to ‘package’ that content for delivery. Thanks to higher density compute devices, low-latency transcoding and high-performance processors, the ability to process content based on the required protocols for each device type and delivery network can now happen at any point in the network. By placing this functionality at the edge rather than in the cloud or a centralized data center, providers can better manage both storage and bandwidth requirements. When combined with real-time analytics performed by compute devices at the edge, the content is packaged more efficiently and delivered more effectively. The edge is able to influence both operating models and revenue models.

However, the real promise of edge computing lies in personalization. The rise of the edge allows content providers, service providers and brands to engage the consumer like never before. By gaining greater, real-time insight into consumer preferences and behaviors, content providers can recommend the most relevant content, service providers can optimize content delivery and brands can dynamically insert compelling ads.

Think about all the data that can be collected from devices such as smartphones, set-top boxes, and dongles. At minimum that data provides insight regarding end-user preferences, location, time of day and viewing habits. Incremental data about weather, social sharing, searches or activity levels can be gleaned from the same or other devices. Simultaneously, data is being collected regarding network performance that could impact the consumer’s quality of experience. Analyzing pertinent data at the edge and taking action in real-time, will lead to an improved and personalized experience.

For example, suppose the end-user commutes by train every day. Data collected would validate the time of day, length of commute, preferred device, network type and content genre. This data then informs both content recommendations and content delivery formats, simplifying the process by which the end-user discovers and enjoys content during their commute. Alternatively, perhaps it’s 3pm on a rainy school day. A TV is turned on with parental controls in place. Recommended content may include longer form content due to parents wanting to entertain children for an afternoon due to the poor weather. Family oriented advertising can be inserted. The data is collected from both the STB and a smart home thermostat.

What’s happening at the edge? Processing power is in demand. Data is being stored, processed, analyzed and acted upon. Functions are being performed to provide more cost effective and efficient service to end-users. Businesses in every industry will be making decisions about where IT infrastructure is located and what functions should be performed in the cloud or in a data center versus those that should be migrated to destinations closer to the end-user. These decisions will be influenced by

• Requirements for real-time action (or, the impact of latency on an action)
• Volume of data being sent back and forth to cloud-located platforms
• Compute requirements (or, can the edge support the level required level of processing)
• Readiness of edge devices and applications (can they actually perform as needed)
• Security

The edge will influence both operations and economics. The general acceptance and adoption of Software-as–a-service business model introduces greater flexibility in how and where functions are performed. The cloud is not dead - it’s just expanding. Some call this Fog computing. It’s all about getting closer to the end-user and serving them better.

What’s your perspective?