Analysis: The economics of edge computing

Analysis: The economics of edge computing
Joe is the chief analyst at Mobile Experts. He focuses on the IoT and 5G vertical markets and leads a team of six analysts to triangulate on accurate market forecasts.

Moving your computing workload to the edge will cost a bit more than you’re accustomed to. Life is good down on the data farm where a hyperscale environment has driven some incredibly low costs. Moving the platform to the edge will set you back, but if it’s implemented properly the cost can be contained.

As shown below, comparing a hyperscale provider’s edge configurations and their regional data centres, we see that the edge platform costs about 35 – 55% more. (We are making a three-year Total Cost of Ownership (TCO) comparison here, including up-front costs and three years of operations.)

The additional cost comes from the higher relative cost of real estate, cooling, maintenance personnel, and other factors.   

The next question that many enterprises will ask: Should I build my own edge cloud, or should I use a platform such as AWS Outpost to set up a cloud instance locally? What’s the most cost-effective approach to get the control and low latency of on-prem cloud?

Mobile Experts has worked with major cloud providers and enterprises to develop a detailed comparison of DIY edge computing vs the cost of a hyperscale provider’s edge cloud solution.    

Based on a three-year TCO analysis, we see some clear cost differences between these two cases.  Our cost modelling reveals some markets that clearly benefit from the use of hyperscaler platforms … and we see some other markets where the DIY approach is still used because a longer time horizon can make a DIY investment more affordable.  

Our MEC cost modelling provides some clarity on the economics of edge computing and the cost premium to enable super-low latency industrial applications. Edge computing will become part of the ingrained vocabulary in industrial automation, retail, healthcare, and warehousing applications as enterprises increasingly look to run their business in real-time on local compute platforms. 

Specifically, we put together one case study in which a factory is outfitted with robotics, video analytics, automated guided vehicles (AGVs), IoT devices, and of course employees with broadband access. We calculated the overall cost for both 5G connectivity and edge computing to handle the entire workload.

Overall, our conclusion was that building a “do it yourself” edge cloud can be roughly 90% more expensive than Microsoft Outpost. This is despite the need to pay for the 24/7 capacity of the Outpost rack, for a factory that doesn’t operate 24/7.

The reason is simple. By taking advantage of the economy of scale from huge hyperscale data centres, the server hardware and software licenses for the Outpost stack can be much cheaper than a customized solution for one company.

We still see companies that develop their own hybrid clouds, (the oil and gas market is a good example, where the need for data control is extreme and the analytics are very customized). But the compelling cost savings of a local edge cloud – based on hyperscale economics – is hard to beat, especially for tasks that are similar to tasks performed in other enterprises.

Expect the trend for “Industry 4.0” to move toward this kind of platform.

(Photo by Sharon McCutcheon on Unsplash)

Interested in hearing industry leaders discuss subjects like this? Attend the co-located 5G Expo, IoT Tech Expo, Blockchain Expo, AI & Big Data Expo, and Cyber Security & Cloud Expo World Series with upcoming events in Silicon Valley, London, and Amsterdam.

Tags: , , , , , , ,

View Comments
Leave a comment

Leave a Reply