Welcome to the new Energy Central — same great community, now with a smoother experience. To login, use your Energy Central email and reset your password.

Virtual Retrofit for Data Centers

Streaming a ‘Sci-Fi’ on Netflix, reordering Amazon AA batteries for the remote or reviewing a Google search for the recipe you made last Sunday…  These and many other tasks are completed with ease for us, the user, but the hardworking data centers supporting cloud-computing services are in need of a virtual retrofit.  Companies have moved to the cloud and with that, cloud data centers have grown in capacity and demand.  Video accounts for 80 percent of internet traffic and data center traffic has grown by 8x with plans for continued growth.  Statistics show that the global cloud computing industry was projected to hit $250.05 billion in 2021 and reach the $791.48 billion mark by the year 2028.   With growth comes demand, the demand for electricity. It is estimated that moderate to large data centers use power densities 100 times that used by a typical office building.  In 2020, data centers accounted for 1% of global energy use.   Data centers that host the components of cloud computing such as servers and air conditioners also release huge amounts of CO2. This has become a growing concern and organizations are looking for ways to reduce consumption and emissions.  

The recent COP26 conference in Glasgow, the U.K. stated that every organization will be required to come up with a plan to reach net zero carbon emissions to avoid warming within 1.5 Celsius. In the states, The US Office of Energy Efficiency & Renewable Energy has provided a wide range of efficiency advice for data centers. In fact, the Department of Energy (DOE) supports other agencies and national laboratories in this effort through its National Data Center Energy Efficiency Information Program.  To reduce traffic and ultimately consumption, companies are innovating their methods and applications.  Software adjustments include load balancing, resource allocation, job scheduling, and Virtual Machine (VM) virtualization and migration.  The use of green codes reduce consumption because the software is written with less instruction for processing and less memory. Computational storage, where data is processed at the storage device level to reduce the amount of data that has to be moved would also reduce energy usage. In existing facilities, data centers have already started consolidating workloads by downsizing servers, storage systems and networking systems.  For new locations, companies are looking to build in cooler climates and therefore reduce the need for cooling systems. Simple adjustments to hardware like low-power CPUs and energy-efficient computer monitors have also been beneficial.

In a poll, some IT decision-makers ranked the cloud as the most overhyped technology.  While the technology may be overstated, energy efficiency for cloud-computing is not.  Last year, Amazon, Microsoft and Google made up more than 50 percent of the world’s largest data centers.  What more can be done to reduce the energy consumed by the on- or off-premise data centers that save our film queues, shopping carts and recipes?