Originally posted to Data Center POST
The time may (finally) be right to talk seriously about high density edge data centers. High density is one of those data center industry predictions raised consistently over the last decade that will enable all of the futurist dreams fueled by the promise of AI, IoT and automated everything else. Regardless of how often this pet prognostication is invoked, it just never seems to happen.
What is so different this time? Frankly, the world has changed and the idea of the data center is changing with it. At the edge, the data center is no longer simply a product seeking to fit any problem to its solution, and companies no longer need to bear the cost on their own. Deployments that have the most impact are those planned as a set of capabilities tailored to serve the growing and varied needs in a specific location.
What Took So Long?
In the enterprise and hyperscale world, the answer was simply space at the location that offered inexpensive land and power. The core was built around sprawling centers with massive connectivity and robust power where more compute simply meant more racks. There was plenty of space, plenty of power and conventional cooling was easy and cheap, all things considered. Basic economics held rack densities constant at or around 5kW, despite the futurist projections.
We see the downside of that path playing out today. We have massive but inefficient data centers where cooling and power max out well before rack capacity. Even in Data Center Alley, we have a hodgepodge of ad-hoc data centers that sprang up to meet exploding needs and driving land prices to astronomical levels. These were the days when edge was still a concept based on micro data center products. But those days are fading and gone.
To read the full article, please click here.