Hyperscale Data Centers: How They’re Impacting the Data Center World

datacenterHawk - A podcast by datacenterHawk

Categories:

Interested in Hyperscale data centers? Sign up for our free hyperscale data center course: https://lp.datacenterhawk.com/hyperscale-business-development-fundamentals Or get a quick 15 minute demo of our platform: https://lp.datacenterhawk.com/request-a-demo?utm_source=youtube&utm_medium=youtube&utm_campaign=demo ––––– It’s no secret that “hyperscale” has been a hot topic in the data center world for a while now, but what impact are they having on the data center industry as a whole? Let’s find out… While a data center can support hundreds of physical servers and several thousands of virtual machines, hyperscale facilities support thousands of physical servers incorporating millions of virtual machines. So yes, hyperscale data centers are much larger in both design and capabilities, but we’re also beginning to see their influence spread beyond their own four walls. With the continued growth of the hyperscale data center market (which, according to Global Market Insights, is expected to rise from $20 billion in 2018 to $65 billion by 2025), every major data center company has had to develop a specific strategy around how they will approach this ever-increasing sector of the market. How Data Center Companies Handle Hyperscale When it comes to leasing wholesale and build-to-suit data center space, hyperscale companies are controlling the demand, making them the primary customer that decides how data centers develop worldwide. Because of the massive growth of cloud service providers like Amazon, these hyperscale companies are not able to self-build all of the capacity their growth requires. So part of the burden rests on colocation providers to help meet the increased demand. We have seen a gradual increase in the amount of capacity hyperscale companies lease vs. build in recent years. In order for colocation providers to capture some of this demand, they need to have a strategy for how to approach land acquisition, development, and a sustainable path to growth. An example of this can be seen on Facebook’s Oregon campus. Originally, they had three 330,000-square-foot data center buildings. But recent improvements to the campus have created room for up to seven 450,000-square-foot buildings. As for creating a path for growth, new campuses being built by data center REITs are being designed with long-term roadmaps to have 350 MWs of capacity or more (up from the typical 50-100 MW). How Hyperscale Affects Data Center Design In the past, data center companies would build a facility that they felt would attract the most number of customers. This led to over-engineered facilities that were not cost-effective. With many “multi-tenant” facilities now going to a single tenant on a pre-lease basis, companies are designing building components like generator redundancy to a single customers' requirements. This ultimately brings the cost of the development down as it removes the guesswork. Additionally, while economics is still a significant factor in site selection, both the cost of the lease and the cost of power, hyperscale companies have become more focused on the scalability factor. Instead of only asking how much this capacity will cost today, hyperscale developers must also think about how much it will cost to increase the power usage in the future. Hyperscale data centers consume massive amounts of power, often creating a large carbon footprint. To offset this issue, we see large hyperscale companies developing their own wind farms and solar energy plants as a means of supporting their operation. Furthermore, indirect evaporative or direct evaporative cooling techniques are used to handle water conservation. These are just some of the questions raised by the development of hyperscale data centers around the world. The impact of hyperscale growth will only continue to increase. In the meantime, stay tuned for the next article in our hyperscale series where we’ll discuss the future of hyperscale data centers.