The Edge or the Cloud? It Depends on the App


There’s no arguing that the cloud has transformed the way organizations deal with data and apps. It has freed IT departments from constantly provisioning and managing storage, while bringing overall costs down. For these reasons and more, companies everywhere are moving more and more data and compute tasks to the cloud every day.

It’s important, however, not to go overboard. No matter how great the cloud is, there are many applications that will simply never be a fit for what the cloud can provide. That doesn’t mean that managing them yourself, on-premise is the way to go, either. In fact, many of these apps require performance and security metrics that also make them a poor fit to be kept in-house.

The solution? Utilizing the edge keeps data, and compute, close to where it’s being used, eliminating some of the limitations of the public cloud. As concepts like the Internet of Things and autonomous devices take shape, the types of apps that are better suited to the edge continue to increase. How do you know if your app will be better served using the edge? Here are three characteristics to keep in mind.

1. Any app that requires machine-to-machine communication

The IoT has introduced us to the concept of devices making decisions and acting on their own. Likewise, smart and autonomous vehicles illustrate the importance of instant, flawless communication between these machines. For any app that requires quick, seamless interaction between devices, with little to no interaction with a person, getting data to a far-off cloud to analyze and then send results and instructions back to the devices simply doesn’t work. Connections for many of these apps have to be instant, and can involve multiple devices at a time.

Only by making use of compute and data near where the devices are working—at the edge—can these applications reach their full potential. And their potential is great. Everything from transportation to manufacturing to healthcare will eventually benefit from autonomous, machine-to-machine interactions. “Eventually” isn’t that far away, either.

2. Any app that can’t tolerate latency

One of the issues with sending data to the cloud to be analyzed is that often cloud data centers are located far away from cities, where a lot of the action is actually taking place. This makes sense, of course; land is less expensive farther away from cities, and this adds significantly to the cost benefits of the cloud.

However, when data travels, it faces a speed limit: the speed of light. At one foot per nanosecond, that speed limit is admittedly pretty fast, but that still means that the speed of data transfer is a limiting factor. When you add in servers, switches, and routers, the time it takes data to travel to a data center hundreds or thousands of miles away, and back, is significant. For financial transactions where milliseconds can mean millions of dollars, or healthcare where any delay is dangerous, the edge is a perfect fit. It virtually eliminates latency due to traveling data, letting those critical apps perform as intended, without any delay.

3. Any app that generates excessive data

Whether apps require zero latency or are autonomous, they have one thing in common—they generate a ton of data. In reality, many apps today generate an amount of data that would have been unthinkable even five years ago. Any app that needs to process this data and act on it rapidly—even if it doesn’t have the strict requirements discussed above—is likely not the best fit for the cloud.

Why is that? Think about how most people use the internet: downloading music, streaming video; until recently, it’s all been about downloads. So, the internet has been optimized for download speed. In fact, most service providers explicitly have slower promised rates for uploads than downloads. As usage patterns shift—uploading photos and videos to social media, for example—it creates a bottleneck when uploading to the cloud.

The edge doesn’t face these restrictions. Uploads and downloads happen with equal speed, eliminating any worry about upload delays for those apps that are generating hundreds of gigabytes of data every day.

The use cases for the public cloud are many, and well-documented. That doesn’t mean the public cloud is the best fit for all your apps and data. So, before you move an app to the cloud, figure out whether it will perform better if you make use of the edge.

Ellen Rubin is CEO and co-founder of ClearSky Data, a Boston-based company that provides enterprise data storage as a fully managed service. Follow @ellen_rubin

Trending on Xconomy