3 Data Center Innovations to Watch

Data centers are essential to the digital economy and they are a remarkable feat of forward-thinking engineering. As a leader in the colocation data center industry, we are continuously focused on driving innovation while delivering to the highest standards of operational and environmental excellence. As an example, our carbon emissions fell 60% between 2015-2019 on an absolute metric tons of carbon dioxide equivalent (mtCO2e) basis, even as our overall data center energy footprint and consumption has nearly doubled. At the same time, our International Business Exchange™ (IBX^®) data centers have an industry-leading track record of 99.9999% average global uptime.

Here are 3 trends we are watching:

1. Sharing and applying AI/ML data is an evolving process: In our initial AI in the data center pilot, one of the key lessons we learned is that lack of quality data is a real challenge. The devices within our data centers are well-maintained, so they don’t fail very often. This means that even with sensors on a large number of devices, we may not have enough data by ourselves to teach the machine learning (ML) model to recognize failure patterns. As in many industries, we believe data sharing and aggregation from multiple sources, including manufacturers and other operators in the industry, is essential to advancing these models. There are different data sharing models that can be leveraged to enable data exchange without compromising competitive advantage or regulatory compliance. One way is federated learning in a neutral data exchange marketplace, where you bring compute to the data and only share out high-level, anonymized aggregated models that don’t identify any particular devices. Through our partnership with Nvidia, our customers can flexibly leverage the same toolsets that we’ve tested to support their own machine learning projects.

The second key insight we had was that it’s really important to focus on the problem you want to solve. While it was very easy to setup ML models that could detect certain abnormalities, such as a pump that stopped working based on its vibration signal, the reality was that we knew that two hours prior to the stoppage because the amount of water going through the pipe was reduced. Thus, the question became, where does it make the most sense to invest in training these models? In a traditional IBX data center, it really boils down to critical equipment – cooling and power. But we are also looking at how it can be applied in non-traditional data centers such as the modular edge data center that was discussed in the blog, “A Quick Guide to Choosing the Right Data Center Design.” In these unmanned facilities, the ability to automatically detect and correct irregular events will be critical, so we believe AI will play an important role there.

2. Power delivery is going virtual: Just as virtualization of servers, networks, and storage has brought benefits such as increased efficiencies, scalability and speed, the virtualization of power delivery is expected to bring a much tighter level of integration between facility and customer infrastructure to support workload orchestration. At Equinix, we’re collaborating with a number of partners on a 21^st century power delivery platform to advance this vision. By integrating next generation power storage technologies such as sodium ion with in-house developed technologies such as internet of things (IoT) enabled intelligent circuit breakers, software controls and a powerful API, we will be able to provide our customers with power-aware workload orchestration. This will open the door to other advantages such as improved utilization of available power, reliability, sustainability and visibility, as well as expanded capacity/density and more automation opportunities.

3. More flexible energy platforms are on the horizon: Energy has always been a major consideration for data centers. One of the first questions asked when selecting a potential location is “where is the electricity?” The challenge of procuring good quality power, combined with the fact that we have a goal of 100% renewable energy, mean that energy is a big focus for us.

In January of this year, we established a new Energy Efficiency Center of Excellence (CoE) that will pilot a globally coordinated approach for maximizing energy efficiency improvements. I’m involved in a number of projects to better understand the energy landscape and how Equinix can remain agile through things like flexible power platforms, local generation, simplifying how we connect to next-generation power sources, energy storage, as well as other conservation measures. Equinix IBX data centers have always been connected to the communities we serve through fiber, networks and, in some cases, through supplying energy back to the grid. But as we look ahead we are also thinking about how emerging technologies and the electricity ecosystem around our data centers might work in the future.

At Equinix, we take innovation seriously

I’m really proud of the data center innovations that Equinix is actively investing in. Our engineering team is developing concepts for building and testing new innovations using a depth of quality thinking and collaboration that I have not seen anywhere else. And the best news is, as these data center innovations are good for Equinix, then they’re going to be good for our customers and good for the planet too.

Learn more about our green design and data center innovations.