Across the globe, higher education institutions are becoming more attentive to climate change. Computing infrastructure will be a part of this conversation at a rapidly increasing rate. Moving on-campus datacenters to a public cloud is one of the ways a college can reduce the carbon footprint of information technology resources. The move to the cloud has been prevalent amongst private companies and some colleges for years now, but higher education as a whole seems to be taking a more cautious approach. Here, we will examine how cloud computing helps IT datacenter operations become more sustainable.
Cloud data centers are built for efficiency! All major cloud service providers understand that wasting energy costs them money. As they design and build data centers, they create the most efficient building, lighting, power conditioning, and cooling systems possible. There is even an ISO standard metric used to measure the efficiency of a data center, Power Usage Effectiveness (PUE). Simply put, PUE is a measure of how much total energy use goes to running computing equipment versus cooling, lighting, and other support systems. There are sites, like data center vendor 42U, that offer calculators and additional information on this measurement.
Now to consider the hardware in the datacenter. The hyperscale datacenters that drive the cloud all want to get the most out of every investment. That means working with suppliers and designers to get as much productive output as possible from the energy input. For example, Microsoft believes they reduce energy consumption by 10% or more from hardware efficiency.
Finally, consider equipment utilization and automatic scaling. When building on-premises infrastructure, even with modern virtualization you need to ensure that you meet peak capacity, even if that level of utilization is rarely required. Most institutions have patterns of increasing and decreasing usage, for example more usage during business hours or during peak registration periods. While some automatic scaling is possible on-premises, it would be nearly impossible to fully utilize these computing resources every hour of the day and night. Higher education workloads often fluctuate in a cyclical way where we know the greatest demand will be during peak registration, which typically only accounts for a few hours per year of total usage. To ensure system availability, we tend to allocate more resources than we expect to need. Cloud providers can host many tenants on a single system and can dynamically manage loads across systems. This broadly means that when some customers are using less resources, others can use more, and they can quickly allocate resources to meet the demand. Increased efficiency and scalability are some of the reasons SIG recommends modernizing your infrastructure, Move and Improve, as opposed to the traditional Lift and Shift migration model.
There may still be overallocation of resources in the cloud but compared to an on-campus facility, it is almost certain to be more efficient, especially when you consider the benefits of improved flexibility as your computing needs change. Counting all factors together, migrating the campus data center to the cloud will result in more efficiency, which translates to less energy usage and a more sustainable operation.
In the last post of this series, we will compare the carbon footprint of the four most prominent cloud infrastructure providers in the higher education arena. Read the next post in this series.