20th July 2020
Locking down our Carbon Footprint
I was sitting in our meeting room in October last year when a customer asked a question that surprised me: "What is Cursor's Carbon Footprint?" The surprising thing was that I didn't have an answer. We had no idea of what our Carbon footprint was, and in our eighteen years of business, no one had ever asked before. In researching our business emissions, this conversation started a journey which was more interesting and complicated than I ever expected.
In principle, calculating Cursor's carbon footprint should be straightforward. Our primary emissions include electricity, business travel and greenhouse gas emissions from our air-con. Secondary emissions include employee commuting, transportation and outsourced activities. Once we gathered the numbers, we discovered that roughly 80 percent of our carbon emissions come from the daily commute. Now that we are working remotely because of Covid-19, these emissions have dropped significantly.
Cursor moved to remote working on the 11th March, ahead of the government announcement on 23rd March. The graph shows how our monthly emissions plummeted as we started working from home. The other reductions in emissions during August and December relate to the summer holidays, when some staff work remotely, and the Christmas holidays, when the office is closed.
The impact of remote working
From these numbers, the best thing we can do to keep carbon emissions low is to continue to work remotely after the threat of Covid-19 has subsided. I'm certainly open to this approach, as I can see the benefits to the business and in particular to our staff. A recent McKinsey report found that eighty percent of people questioned said that they enjoyed working from home. Forty-one per cent say that they are more productive now than they were before, and twenty-eight per cent say that they are as productive at home as in the office. We've had people who commute via car or train, from Nottingham and near Grantham or Newark. By removing unproductive travel time, we give employees greater flexibility in balancing their personal and professional lives. Currently, I have no plans to remove the office altogether, although it's likely to shift to more of a hub than the primary workplace, with staff spending one or two days a week together and then the rest of the time working remotely.
Virtual meetings have also been a success throughout the lockdown and will be something we look to continue in the long term. Once people have sorted the usual gremlins with audio and lighting, I've found frequent group video meetings a refreshing change to long face-to-face discussions. I've been able to fit a day's meetings into just the morning, which frees the afternoons for concentrated work without interruptions. Another benefit is when you can invite team members to a meeting with an expectation that they will continue work while keeping an ear to the discussions, and joining in only when needed. Virtual meetings are certainly here to stay, increasing productivity as well as reducing travel and associated emissions.
It's strange to me that even though travel is our main source of emissions, it is classified as secondary emissions so often ignored and not reported. This made me wonder what other indirect emissions are hiding below the surface? We might not have a large supply chain with lots of raw materials, but I've often heard the statistic which says that if the Internet were a country, it would rank as the fifth largest for energy consumption. Surely this should be factored into our Carbon calculations?
The Internet’s Carbon Emissions
My research started with the Click Clean 2017 report produced by Greenpeace, which highlighted an interesting issue. The Internet's energy footprint is going to continue rising, fueled by both an increase in our consumption of data and the spread of the digital age to more of the world's population. Then how we power our digital infrastructure will become a central question in the fight against dangerous climate change.
If more data centres and digital infrastructures use one hundred per cent renewable energy, then our increasing use of digital services could accelerate us towards a renewable-powered economy. However, if the growth of the Internet increases the demand for electricity from coal and other dirty fuels, then the opposite could be the case.
The Greenpeace report finds that Apple, Google and Facebook continue to lead the sector in the use of renewables, energy transparency, energy policy and efficiency, each gaining an 'A' rating. Amazon, however, only manages a 'C' grade for use of renewables, with thirty per cent of their electricity coming from coal (2016), and an 'F' rating for energy transparency.
Amazon Web Services (AWS) hosts some of the world's most popular websites, such as Netflix, Pinterest, Etsy, LinkedIn, Spotify, and Reddit, so Amazon really needs to up its game. Netflix alone is responsible for an estimated fifteen per cent of all Internet traffic, and added sixteen million new subscribers since the start of lockdown alone. Any improvements AWS can make to its use of renewable energy will multiply as traffic to these popular sites increases.
As a business, we provide hosting for about eighty websites, from low traffic websites for startups to large sites for international firms, hosted in data centres in York, London, Dublin and Virginia, USA. I wondered how our technology partners were sourcing their electricity and if it was from clean sources as I hoped, or dirty sources as I feared. I sent the following email to our hosting partners on the 29th April;
"Hello, we are doing a project to work out our total carbon footprint and wonder what information you can provide us about our servers. Are you able to provide information on energy usage for our servers by month?
Alternatively, can you provide server bandwidth total by month which we could then use to estimate energy usage? Can you provide information about how you source your energy? What percentage comes from clean/renewable sources? I would appreciate any information you can provide. Thank you."
So how did they do?
The team at Bytemark encouragingly responded later the same day;
"Thanks for getting in touch with us about this very important topic, one I believe everyone now is (or should be) beginning to take more seriously about their responsibility to the environment and carbon footprint. You have raised some very good questions which I don't, unfortunately, have the answers to within the support team to hand, however, I will make some inquiries and see if I can find what you are looking towards or connect you with the right person with that detail. I will follow up once I can provide you with more information."
I followed up on 1st May, and they said "Just trying to figure out who the best person would be to ask on our side about this. Might take me a few days to get this info/person found for you.".
Then when I followed up on the 11th May, the reply was "I am going to speak with one of our operations team who handles the data centre operations to see if they can help with this detail. Sorry again for the delay in getting this information."
Then finally on the 18th May "I am going to speak with my Team Leader regarding this to avoid tickets being automatically closed while still in progress. Please accept my extended apologies for the delay in this request. I have asked the relevant member of our team for more details, and I am waiting for a reply from them for you. I will send a reminder this morning regarding this."
As of July 2020, we still haven't received any information from Bytemark, and our support tickets have been closed. Total mark: F. No data, all talk no action.
Media temple came back after a day with the following response "Unfortunately, after checking with multiple departments this is not information that we provide or have readily available so that I can even give estimates. Thank you for choosing (mt) Media Temple for your hosting needs."
This reply annoyed me, so I decided to take it up a gear.
"Hello, I am disappointed to hear that Media Temple is unable to provide even the most basic environmental or carbon emission information. I have therefore tweeted your reply and hope to get a better response from Media temple in the face of public criticism."
I didn't hear back from the support team, but my tweet got the attention of Matt from the Customer Success Department who replied "I recently reviewed your request about your total carbon footprint. I am sorry that the agent working with your ticket could not get you
information about the amount of energy usage of your services. I was able to speak with our network work admins at our datacenter about your concern. Your services use the following amount of electricity: 0.005657998424kwh.
I hope that this information was helpful - unfortunately, this information is not readily available to our frontline support team. I will speak directly to our (mt) Leadership Team to see if we can get some information about energy data/carbon footprint on our website."
Okay, a bit of a shaky start but we got somewhere in the end. My Macbook Pro uses much more energy, around 0.061kWh, so that's interesting. Still, nothing about the percentage of renewable energy but after a bit of investigation, it seems like the area of Virginia known as 'Data Center Alley' is served exclusively by a company called Dominion Energy which generates the majority of its energy through gas and coal power stations. So, better details but still no renewable energy. I'd give this a total mark of C.
Things didn't start very well with an automatic response which said "We are currently experiencing a surge in support ticket volume, which is causing longer-than-average response times. We appreciate your patience and understanding, and we look forward to serving you as soon as we can."
However, then on the 4th May, I got the following reply "Thank you so much for reaching out about this. DigitalOcean uses commercial data centre providers to house its infrastructure. Each of these providers is committed to sustainability and seeks compliance with rigorous energy management, green building and environmental standards.".
Now, this isn't the energy usage that I asked for, but then with a bit of digging, I was able to work out that Digital Ocean's LON1 data centre within
the Equinix facility in Slough. This data centre has been powered by 100% renewable energy since 2017 through supplier REGO-backed renewable energy certificates. There's even a website about Equinix's sustainability program including information about carbon reduction programmes, how they build green data centers, a full 2019 sustainability report,
Still, no information about the energy usage of each VM we host with Digital Ocean (and we have a few now), but this has been the best response so far, with a clear commitment to sustainability and information readily available. I'd give this a B+, much better.
Based on this research, I will commit to moving all our hosting to data centres powered by 100% renewable energy within the next 12 months.
The carbon footprint of data transit
I think I now have a much better idea of our carbon footprint across office emissions, transport and data centres. I'd assumed that data centres would make up a significant proportion of emissions, but this doesn't appear to be the case. We've talked previously about the carbon cost of commuting to work, but perhaps there is another type of transport to consider.
The Internet is made from much more than data centres. It's a global collection of computers perpetually moving information in the form of bits across fibre-optic cables and radio signals.
A request for a web page hosted on our Media Temple server starts in our office in Lincoln, jumps to a facility in Peterborough, then hops around London before hitting New York, Washington DC, Portland (Oregon), Scottsdale (Arizona), Culver City (California) before finally hitting our server in Virginia. The server processes the request, and the HTML page, along with all the pictures and associated files begin the return journey back to my laptop in Lincoln.
All this infrastructure requires massive amounts of power, but as each request for a web page can take a different route, it is impossible to calculate exact figures. We can, however, follow the methodology on websitecarbon.com which takes the overall energy use from a report on Global Electricity Usage of Communication Technology then subtracts the energy of manufacturing to get the total energy use of the Internet. Divide this by the complete annual data transfer over the web as reported in the Nature article, How to stop data centres gobbling up the world's electricity to give us a figure of 1.8kWh/GB.
Then if we look at the international average carbon intensity of grid electricity as reported by the International Energy Agency (475 grams CO2e per kWh) and by Ecotricity renewable wind energy (33.4 grams CO2e per kWh). I'm going to assume that 80% of Internet infrastructure is powered by grid energy with the rest coming from renewables which makes the combined total 386.7 grams of CO2e per kWh.
I then looked at our firewall logs to work out the total amount of data we transfer a month (on average 64 Gb/month). Using all these figures we can estimate our data transfer carbon emissions per year to be
The total amount of data transferred (3328GB) * 1.8 = 5990.4 kWh
5990.4 * 386.7 (grams of CO2e per kWh) = 2316.49 kg of CO2e per year
That puts data usage as our second highest source of carbon emissions at roughly 16% of total emissions.
Final thoughts
I’m amazed that data accounts for more emissions than office electricity, something I would never have expected when this journey began. It makes me think of the responsibilities we have as designers to create websites that keep bandwidth low to save on emissions. The Internet is full of automatically starting videos and infinite scrolling content which is unnecessarily consuming data. This bandwidth-heavy content might be popular, but perhaps we should prioritise the environment first and think twice before we put it on our company's homepage.
In this blog post I've been looking at our carbon footprint at work, but at home, we watch our fair share of Netflix, Disney + and YouTube, consuming almost ten times more data than the business does. Add to this all the carbon associated with food, transport and heating means the majority of my carbon emissions come from my home life rather than work. It seems that once again, when it comes to locking down my carbon footprint, I'm looking in the wrong place.
More great reads from Cursor Curated:
