Edited by Christopher Douce, Thursday, 24 Oct 2024, 12:01
On 23 October 2024 I attended something called a Go Green Workshop which was all about ‘Dirty Digital Habits’. Having recently supported a research intern who had been exploring what is meant by green computing, the topic of sustainable computing (and all that means) has become of increasing interest. The event had an alternative title: Go Green Spotlight - Digital Carbon Footprints. Apparently, this month is ‘cyber awareness month’. A lot of the session was facilitated by David Crews, our Digital Services Sustainability lead, with input from other colleagues.
A number of topics were covered in the hour long session. These were, broadly: challenges, current efforts, what we can do, and something about a search engine called Ecosia. I share a bunch of reflections at the end.
Challenges
The full title of this topic was ‘challenges in reducing digital related CO2 emissions’. There was a passing reference to a Guardian article about data centres but there was also a passing comments that data centres would consume 8% of electricity by 2030. This suggested figure is likely to be under threat due to increases in cloud computing, and AI. I noted another challenging comment that was shared, which was: by 2025, everyone will have an interaction with a data centre every 18 seconds. I have no idea where this striking estimate comes from, but I now need to find out.
Current efforts
The full title of this section was ‘current efforts in digital services to reduce CO2 emissions’. In other words, what was the university doing? More specifically, what is ‘digital services’ doing?
During this bit of the workshop, I noted down a few points. One continual action that was being taken was addressing of legacy systems. There was a reference to the scheduled replacement of hardware, with the suggestion that newer hardware means better energy efficiency (I did later ask a question that asked whether the carbon cost of equipment manufacture, and disposal was also taken account of).
There was also a reference to a movement of services from university data centres to data centres managed by third party suppliers (two different vendors were mentioned). On this point, I also noted down the words “a cloud data centre is more efficient than anything that an organisation can run themselves”. The point is, of course, that that energy costs of our own data centres are likely to be higher than the costs of cloud suppliers.
A number of familiar topics were mentioned, such as the movement to different processor architectures on servers (such as ARM processors, the kind of chips that you find on your mobile phone), the use of virtual machines to make optimal use of hardware, and the importance of having your data (and servers) as close to the users as possible. Closeness improves latency (performance) and reduces energy consumption.
On a related note, I was introduced the PUE measurement, which is an abbreviation for Power Usage Effectiveness (Wikipedia). The lower the number, the more efficient your data centre is. I need to look into this in a bit more depth.
What can we do?
Under this heading, I combined notes from two different parts of the presentation which had the titles ‘how we can contribute at home’ and ‘top categories to reduce your digital carbon footprint’. Some of the suggestions were familiar: use energy efficient peripherals, power things off, go paperless, stop hoarding data (particularly on the cloud), and ask ourselves whether we really need to use super high definition data when streaming film. In the work environment, when having team calls, can we get away with only using audio? This obviously reduces how much data we use. Also, use browser bookmarks rather than searching for a website address using search engines. An interesting statistic was shared: a search using Google will use 14g of CO2, a kettle takes 15g (but I don’t have the reference from where this comes from; I need to find it).
An interesting question was highlighted: what is the carbon cost of email? It depends, of course, on the length and the content of an email message. A practical suggestion was to minimise the use of attachments. A further interesting statistic was that 71% of colleagues make use of email attachments rather than linking to files. Shipping files across and between email accounts makes unnecessary use of data storage. A file sent through email can use 50g of CO2 (again, I don’t know where this statistic comes from). Sharing a link rather than an attachment can mean a 90% reduction. Of course, the choice of attachment vs link very much depends on whether we have continual access to the cloud services that we need to use.
Ecosia: Search engine pilot
Towards the end of the session, a search engine called Ecosia was mentioned. More information about Ecosia is available through a useful Wikipedia summary. The reason for mentioning it is that there is a plan to pilot its use as the default search engine.
Reflections
I feel as if I must have missed something. I had never heard of an OU unit called ‘digital services’ before; I’ve always known it as IT (in the same way that I know People Services as HR).
This wasn’t an academic event, and it was less of a workshop and more of a presentation (since the time allocated to discussion was quite limited). I did feel that there was quite a need to discuss digital practices. Also, wearing an academic hat, I have an intrinsic need to look for the sources that relate to the statistics that were shared.
I appreciated the presentation about the carbon cost of email. This discussion reflects a similar question I have, which is: what is the carbon cost of taking a digital photograph? There probably isn’t a simple answer: it would depend on how that photograph is used (and how it is shared).
Given that energy consumption is now important to business units such as digital services, I did wonder about whether there will become a time where organisations may be required to report their energy use. To what extent may statutory reporting be helpful or appropriate? Is this desirable or useful? There’s that old engineering adage of if you can measure something, you can then control it.
Within all this, there’s the need for balance. There’s the need to balance the need to minimise using digital bandwidth with the importance of maintaining and fostering personal connections. There’s also the need to use additional layers of software to maintain security and integrity of systems (such as virus checkers and intruder detection systems), whilst minimising a software stack to reduce server costs. There’s the balance of necessarily carrying out searches to attempt to solve a problem, contrasted with the inherent carbon cost of carrying out searches.
One of the points that I asked about was the physical (environmental) cost of changing hardware. Electronics contain a huge amount of embedded carbon, which comes its design and manufacture. How is electronic hardware safely and responsibly disposed? Also, how to we make the most effective ‘carbon’ decisions about to migrate from one hardware platform to another? Should this come under the remit of software engineering? Are there some standards or guidelines that can be help with decision making?
Tone of this event was very positive. Perhaps it’s my state of mind, but this positivity doesn’t reflect my own understanding of the term ‘climate emergency’. It’s important to consider how we consume digital resources, just as we consume physical resources. Individuals can, of course make a difference, but if individual decisions are not measured (in the way that the use of the Ecosia search engine may be measured), then it’s hard to measure what difference we may have made.
I’m going to have to search for some references.
Many thanks to everyone in Digital Services for facilitating a thought provoking event.
Go Green Spotlight
On 23 October 2024 I attended something called a Go Green Workshop which was all about ‘Dirty Digital Habits’. Having recently supported a research intern who had been exploring what is meant by green computing, the topic of sustainable computing (and all that means) has become of increasing interest. The event had an alternative title: Go Green Spotlight - Digital Carbon Footprints. Apparently, this month is ‘cyber awareness month’. A lot of the session was facilitated by David Crews, our Digital Services Sustainability lead, with input from other colleagues.
A number of topics were covered in the hour long session. These were, broadly: challenges, current efforts, what we can do, and something about a search engine called Ecosia. I share a bunch of reflections at the end.
Challenges
The full title of this topic was ‘challenges in reducing digital related CO2 emissions’. There was a passing reference to a Guardian article about data centres but there was also a passing comments that data centres would consume 8% of electricity by 2030. This suggested figure is likely to be under threat due to increases in cloud computing, and AI. I noted another challenging comment that was shared, which was: by 2025, everyone will have an interaction with a data centre every 18 seconds. I have no idea where this striking estimate comes from, but I now need to find out.
Current efforts
The full title of this section was ‘current efforts in digital services to reduce CO2 emissions’. In other words, what was the university doing? More specifically, what is ‘digital services’ doing?
During this bit of the workshop, I noted down a few points. One continual action that was being taken was addressing of legacy systems. There was a reference to the scheduled replacement of hardware, with the suggestion that newer hardware means better energy efficiency (I did later ask a question that asked whether the carbon cost of equipment manufacture, and disposal was also taken account of).
There was also a reference to a movement of services from university data centres to data centres managed by third party suppliers (two different vendors were mentioned). On this point, I also noted down the words “a cloud data centre is more efficient than anything that an organisation can run themselves”. The point is, of course, that that energy costs of our own data centres are likely to be higher than the costs of cloud suppliers.
A number of familiar topics were mentioned, such as the movement to different processor architectures on servers (such as ARM processors, the kind of chips that you find on your mobile phone), the use of virtual machines to make optimal use of hardware, and the importance of having your data (and servers) as close to the users as possible. Closeness improves latency (performance) and reduces energy consumption.
On a related note, I was introduced the PUE measurement, which is an abbreviation for Power Usage Effectiveness (Wikipedia). The lower the number, the more efficient your data centre is. I need to look into this in a bit more depth.
What can we do?
Under this heading, I combined notes from two different parts of the presentation which had the titles ‘how we can contribute at home’ and ‘top categories to reduce your digital carbon footprint’. Some of the suggestions were familiar: use energy efficient peripherals, power things off, go paperless, stop hoarding data (particularly on the cloud), and ask ourselves whether we really need to use super high definition data when streaming film. In the work environment, when having team calls, can we get away with only using audio? This obviously reduces how much data we use. Also, use browser bookmarks rather than searching for a website address using search engines. An interesting statistic was shared: a search using Google will use 14g of CO2, a kettle takes 15g (but I don’t have the reference from where this comes from; I need to find it).
An interesting question was highlighted: what is the carbon cost of email? It depends, of course, on the length and the content of an email message. A practical suggestion was to minimise the use of attachments. A further interesting statistic was that 71% of colleagues make use of email attachments rather than linking to files. Shipping files across and between email accounts makes unnecessary use of data storage. A file sent through email can use 50g of CO2 (again, I don’t know where this statistic comes from). Sharing a link rather than an attachment can mean a 90% reduction. Of course, the choice of attachment vs link very much depends on whether we have continual access to the cloud services that we need to use.
Ecosia: Search engine pilot
Towards the end of the session, a search engine called Ecosia was mentioned. More information about Ecosia is available through a useful Wikipedia summary. The reason for mentioning it is that there is a plan to pilot its use as the default search engine.
Reflections
I feel as if I must have missed something. I had never heard of an OU unit called ‘digital services’ before; I’ve always known it as IT (in the same way that I know People Services as HR).
This wasn’t an academic event, and it was less of a workshop and more of a presentation (since the time allocated to discussion was quite limited). I did feel that there was quite a need to discuss digital practices. Also, wearing an academic hat, I have an intrinsic need to look for the sources that relate to the statistics that were shared.
I appreciated the presentation about the carbon cost of email. This discussion reflects a similar question I have, which is: what is the carbon cost of taking a digital photograph? There probably isn’t a simple answer: it would depend on how that photograph is used (and how it is shared).
Given that energy consumption is now important to business units such as digital services, I did wonder about whether there will become a time where organisations may be required to report their energy use. To what extent may statutory reporting be helpful or appropriate? Is this desirable or useful? There’s that old engineering adage of if you can measure something, you can then control it.
Within all this, there’s the need for balance. There’s the need to balance the need to minimise using digital bandwidth with the importance of maintaining and fostering personal connections. There’s also the need to use additional layers of software to maintain security and integrity of systems (such as virus checkers and intruder detection systems), whilst minimising a software stack to reduce server costs. There’s the balance of necessarily carrying out searches to attempt to solve a problem, contrasted with the inherent carbon cost of carrying out searches.
One of the points that I asked about was the physical (environmental) cost of changing hardware. Electronics contain a huge amount of embedded carbon, which comes its design and manufacture. How is electronic hardware safely and responsibly disposed? Also, how to we make the most effective ‘carbon’ decisions about to migrate from one hardware platform to another? Should this come under the remit of software engineering? Are there some standards or guidelines that can be help with decision making?
Tone of this event was very positive. Perhaps it’s my state of mind, but this positivity doesn’t reflect my own understanding of the term ‘climate emergency’. It’s important to consider how we consume digital resources, just as we consume physical resources. Individuals can, of course make a difference, but if individual decisions are not measured (in the way that the use of the Ecosia search engine may be measured), then it’s hard to measure what difference we may have made.
I’m going to have to search for some references.
Many thanks to everyone in Digital Services for facilitating a thought provoking event.