When most people picture corporate contributors to climate change, they likely imagine the fossil fuel or transportation industries. But many are starting to recognize the environmental impact of a more hidden industry: computing.
“The computer industry is like a teenager; it needs to grow up and become responsible,” says Bill Hoffman.
Hoffman is the CEO and chairman of Responsible Computing, a new consortium that aims to expand sustainable computing practices. Launched on May 20 with founding members Dell and IBM, Responsible Computing is asking the industry to commit to being “responsible tenants of planet Earth.”
It’s a goal all engineers should aspire to. Thankfully, there are easy ways that each of us can compute more responsibly.
Computing Is Polluting
Today, the information and communications technology industry is responsible for generating between 1.8 percent and 3.9 percent of greenhouse gas emissions globally, according to IBM. Data centers alone account for 3 percent of annual energy consumption.
These numbers aren’t exactly shocking. Hoffman has been in the computer industry for over 40 years, and in all that time, he says he has never encountered companies that cared about the efficiency of their code, their use of electricity, or their greenhouse gas production.
“Now, we know we can do better,” Hoffman says. “[Responsible Computing] is a holistic approach to get companies to critically evaluate metrics that impact their computing practices.”
How to Compute Responsibly
The new consortium defines responsible computing as “a systemic approach to addressing current and future problems in computing.”
“It’s a framework that will help to guide the change that is required across industries,” clarifies Marc Peters, CTO for Deutsche Telekom in Cologne and a distinguished engineer with IBM.
There are six domains that make up that framework, as outlined by Responsible Computing: data centers, infrastructure, code, data usage, systems, and impact.
The consortium intends to generate use cases and best practices for responsible computing. It will initially use a self-assessment sustainability maturing model developed by IBM for consulting clients. If all goes to plan, this will lead to specific key performance indicators (KPIs) and success metrics that the industry can adopt.
But there’s no need to wait—every engineering company can immediately begin the journey to more responsible computing.
Six Steps to Responsible Computing
Evaluate Your Current State of Computing
The first step in introducing sustainable computing practices is to evaluate your current hardware and software usage and identify key areas for improvement. By assessing where you think energy is wasted, you can focus your time and resources on making realistic improvements as quickly as possible.
Peters says it is crucial to determine your level of sustainable maturity, so that your goals can be focused and specific. You could also consider metrics beyond sustainability, such as by identifying and eliminating bias in any systems.
Improve Code Efficiency
Code is often written without significant regard for its efficiency. However, improving code to run with fewer machine cycles reduces energy usage (and often results in more robust code). The more often a piece of code is executed, the bigger the difference a slight optimization can make.
Additionally, making code open source can help other organizations achieve similar tasks without reinventing the wheel—a core tenet of the Responsible Computing consortium.
Reduce Dark Data Production
When asked about areas where companies can immediately focus their sustainability efforts, Hoffman highlighted the dark data problem.
“What bothers me is dark data,” says Hoffman. “Data gathered all day, every day in the hope that it’ll eventually be useful, but you just end up paying to collect and store it; nothing meaningful ever happens.”
If your organization collects this type of data, take a moment to reconsider it. Is the data proving useful? Do you need to store all of this data, and for how long?
Reevaluate Data Center Setups
Data centers are one of the most significant drivers of energy consumption in the computing sector. Thankfully, some straightforward steps can be taken to reduce energy usage in data centers and server farms.
First, set up hot and cold aisles to optimize heating and cooling, which can reduce energy usage for temperature regulation. Second, integrate cycles of downtime where servers are shut down when they are not in use. For example, schedule blocks of time for specific activities to ensure that equipment is running only when necessary and parallelize tasks to reduce energy usage.
For example, Facebook uses a load-balancing system called Autoscale that reduces the number of servers the company needs during low-traffic hours. In 2014, Facebook reported that Autoscale achieved power savings of about 10 to 15 percent.
Improve Hardware Efficiency
As new equipment is purchased or replaced, companies can consider sustainability by selecting more energy-efficient processors. When scaled to many chips, even small differences add up to significantly reduce energy usage and operating costs.
Even if you aren’t using on-premises solutions, you can still consider efficiency. The Green 500 is a list of supercomputers ranked by energy efficiency, several of which are publicly available for use by researchers or other NGOs requiring high-performance computing (HPC) solutions. The top spot currently belongs to the Frontier Test & Development System at Oak Ridge National Laboratory (ORNL) in the U.S., a smaller version of the full Frontier system, which ranked number two. The Frontier system also holds the top spot as the most powerful supercomputer in the world, proving that top-of-the-line computing can still align with sustainability metrics.
Switch to Renewable Energy
Although this one is a no-brainer, many data centers and IT companies rely not only on fossil fuel-driven grid energy but also on diesel-run backup generators.
Using solar panels or wind turbines to support disruptions to the grid can greatly reduce a company’s reliance on fossil fuels without sacrificing reliable energy sources. Tech giants are already making these changes, with Amazon, Google, Facebook and Microsoft among the largest corporate purchasers of renewable energy globally.
The excess heat generated by data centers can also be used as an energy source. An IBM data center in Switzerland currently uses its heat to warm a swimming pool, and IBM is working on a project that will recycle excess heat to power cooling devices within its centers.
Where Do We Go from Here?
Without a set of industry standards, it can be difficult for companies, employees and customers to define a truly “sustainable” technology enterprise. The Responsible Computing consortium aims to provide them with the community and KPIs to facilitate the necessary changes that will limit the environmental impact of computing.
With the climate crisis creating an existential threat to society, it can sometimes be challenging to grasp the urgency with which some of these changes need to be made. Peters and Hoffman hope the Responsible Computing consortium will grow over the next few years to light the way for organizations seeking more sustainability.
“If a company is using technology, there is a play for them here,” says Hoffman.
Within six months, Peters expects the first group of consortium members to provide data on how membership has supported their transition to responsible computing. In addition to the cofounders Dell and IBM, consortium members include industry stakeholders like Slingshot Simulations and research institutions at Purdue University and the University of Edinburgh.
Interested organizations can join the consortium and sign its manifesto, which includes principles related to sustainability, openness, accountability and inclusivity, among others.
But whether you join the consortium or not, integrating sustainability as a KPI seems to be a crucial step for the future of engineering and technology.