Datacenter Trends

Death of the Datacenter: Q&A with GIGA Chief Jake Ring

Does the rise of the public cloud mean the end of the enterprise datacenter? That's an oversimplification of a complex issue, according to one CEO whose company specializes in hyperscale, modular datacenters.

• By John K. Waters
• 10/16/2019

Earlier this year, Gartner analysts David Cappuccio and Ross Winser published a report called "Top 10 Emerging Trends Affecting Digital Infrastructure and Operations in 2019." Their list included, unsurprisingly, serverless computing, artificial intelligence (AI), network agility and edge computing.

But one of their top trends, "The Data Center is Dead," did seem to be a bold prediction to more than a few people, including Jake Ring (the tech exec, not the baseball player). Ring is the co-founder and CEO of Atlanta-based datacenter provider GIGA Data Centers. He's also a speaker and blogger, and he recently wrote a different take on the analysts' conclusion that the dramatic growth of the public cloud coupled with the high number of enterprise facilities going up for sale are the harbingers of the death of the enterprise datacenter.

I caught up with Ring when he was in the Bay Area a while back to lead a session at a datacenter conference. I asked him to say a bit more about his conclusion that, to paraphrase Mr. Twain, reports of the enterprise datacenter's death are greatly exaggerated.

Waters: In your post, you wrote, "There is no disputing that the adoption of public and private clouds, as well as colocation hosting, are all growing at enormous rates," and, "the number of enterprise datacenters up for sale is greater than ever." Why isn't that combination a sign that enterprise-owned datacenters are on their way out?
Ring: I do agree that the move to the cloud and virtual environments is going to accelerate. Companies are always looking for ways to reduce costs, and it's IT's turn on the chopping block. But, as I pointed out in that article, the cloud providers can't build enough capacity fast enough to satisfy current needs, let alone the growth in demand the analysts are predicting over the next few years.

You cite market growth forecasts in your post from three analyst firms. IDC has said global public cloud revenue growth will reach $277 billion by 2021. Forrester says it'll reach $323 billion by 2021. And Gartner says it'll reach $278 billion.
Right, but not all workloads are meant for a public cloud of shared resources, and some enterprises have investments in IT hardware that are strategically valuable for them to own and manage.

In fact, you also cite a 451 Research estimate that in 2018, private cloud revenue was about 43 percent greater than public cloud revenue and growing only 2 percent less quickly.
Privately hosted clouds still account for a bigger share of revenue than public clouds. And private clouds live within enterprise or colocation datacenters, not in the shared resources that reside in hyperscale facilities.

But the trend is still away from enterprise-owned datacenters, right?
What the Gartner analysts were saying is if you're an enterprise client, you should not be operating your own datacenter. You should be getting out, unburdening yourself. You should be hosting with either a public or private cloud configuration, or moving into colo, if your CFO wants you to continue to have something to depreciate or you need to keep your equipment.

What's the problem with that recommendation?
Nothing, but again, although the hyperscale companies continue to build sites, they don't have enough capacity to support this predicted growth on their own. They have, in fact, been increasingly turning to colocation providers for help.

So, great news for the colocation market?
Sure, but the basic problem remains the same: capacity versus demand. Regardless of which forecast you use, a huge capacity of colocation datacenters is needed over the next three years. To make sure the infrastructure growth is capable of meeting the analysts' predicted requirements for all workload processing needs, future datacenter retrofits and new builds have to be low-cost and highly efficient. And they have to support higher power densities and have the ability to rapidly scale.

One strategy for dealing with this problem -- the solution your company provides -- is the use of prefabricated modular technology.
The modular approach allows us to bring hyperscale performance down to the retail colo. Our approach is to build out multitenant colocation facilities that allow you to add capacity in a modular way. Think of Lego bricks. You need more capacity; we just add more enclosures out on the floor. It only takes us two weeks to assemble one onsite. We can take an empty warehouse and convert it into a datacenter in as little six months.

That's basically what you did in your first datacenter in Charlotte, N.C., with your WindChill modules, which are your Legos.
It's actually in a suburb of Charlotte called Mooresville, which is the local home of NASCAR. In fact, the facility was the former Dale Earnhardt Motorsport Manufacturing site. We bought the facility in September of last year, and we were done with all the construction, networking -- everything -- in May. And that's with some delays.

Talk about the modules.
It's a proprietary design that's similar conceptually to the ITPAC [IT Pre-Assembled Component] modular design used by Microsoft. It uses fresh air to cool the servers, using an adiabatic cooling system -- the air passes over water flowing through evaporative cooling media. The design uses negative air pressure to pull the air through the servers and into a contained hot aisle, which prevents mixing hot and cold air.

Do I have it right that you consider your company to be a next-gen datacenter provider?
We do. There are a lot of datacenters out there still using the raised-floor approach. They've been around since '86, and when they add capacity, they're just building out more raised-floor capacity. So they're going to be around for a while yet. But the inherent design limitations of how much cooling you're going to be able to provide to racks in a row are costing people money. There really is a better way to do it.