Earth Notes: SPAM's Carbon Footprint @Home (2017)

Updated 2021-07-12 17:45 GMT.
The carbon cost of fending off SPAM (and other networked malice) at home...
I run my primary Internet servers at home, including my main mail server and the Web site that serves this page for example. Noticeable effort and energy goes into fighting off attempts to SPAM me (~10,000 email attempts per day on this server) and to hack me.

(Back story: from the moment that my company connected to the 'live' Internet in the early '90s as one of the UK's first ISPs, we have run a firewall. Thus we have paid an energy cost to defend ourselves. Indeed, almost the very first packets I logged on our shiny new live connection were attempts to hack or SPAM us, in those days from South America...)

Hosted Servers

SPAM logging on console

I use plenty of services hosted outside my home, such as DNS, mail and Web alternatives/secondaries. Never mind the entire network infrastructure that is the Internet. I wondered if I could estimate how much of my carbon footprint at home could be attributed to fighting off unsolicited and unwanted offers of sexy photos, guaranteed penis pills, how to MAKE MONEY FAST, power my house for free forever for only $49, and even how to buy cheap banner ads in the Daily Mail. Oh dear.

A slight complication is that my server is powered year-round off-grid. When there's anough left over even my Internet connection FTTC box and router are pulled off-grid too. And my laptop can run off-grid, manually.

Even when back on grid, my home grid-tie solar PV system generates about double our house's actual electricity consumption. Thus covering the carbon for that and the natural gas that still heats our home and our hot water.

Also, the carbon cost of keeping out the bad guys isn't just in my router and firewall and wasted network bandwidth. Nor does it stop at the fact that none of the CPUs involved get to idle properly or for long enough by being forever woken by SPAMmers' packets, for example. Also, SPAM that makes it past those defences that I have to waste brainpower on uses energy from food, which is some of the most carbon intense going. Energy efficiency is hard when your servers are being prodded by the bad guys at least every minute for 20+ years!

Let's keep things simple and pretend that the carbon cost of the power to run my services is something like the GB grid's probable 2017-05 ~400gCO2e/kWh and falling. Also that no more than that is fighting SPAM etc at home. (My local solar PV energy might otherwise have displaced carbon elsewhere on the grid.)

I'll also ignore the climate change impact of my lentil-powered fury when SPAM gets through all of that anyway...

Direct Energy Costs

The obvious direct energy costs per day at my house are:

So a little under 0.5kWh/day or ~200gCO2e/day or ~73kgCO2e/y tops.

Maybe it would be fair to attribute a few kg of CO2 per year directly to the on-going costs of fighting SPAM (OPEX) for me at home. That would ignore the amortised enormous infrastructure effort and complication (CAPEX) to design and maintain good filters for my mail client's filters, sendmail rules, SPAMHAUS RBL, Linux OS security, yadda, yadda, ...

Maybe several kg per day of CO2 could be attributed to energy wasted in the rest of the Internet in attempts to deliver bad traffic to me, eg 0.3g per attempted SPAM times 10k attempts per day.

(Footnote: today while writing this I seem to be getting a lot of phishing SPAM with poisoned but plausible links or attachments to open, and that are getting through the various levels of filtering. That really does consume expensive mental effort and network bandwidth.)

Not Just SPAM

Somewhere the phishing blurs into low-level DDoS harassment and slow continuous break-in attempts and skiddies unable to rite code rite. As of 2017-10 I've had to adjust my stats gathering and add a filtering rule or three to defang 25% completely bogus mindless repeated downloading of one page from this site. A significant extra cause of CPU demand spikes on the server is clusters of ssh bruteforce break-in attempts... All this consumes energy (mine, and stuff then needing to be generated by gas-fired plant somewhere on the grid ultimately), and adds to CO2 emissions. If only the ultimate perps could be made to pay for the climate clean-up...

20s "SPAM log" (poster) Uploaded . Download MP4 clip 2277kB (i) (may not be available in data-saver/lite mode), download low-bandwidth MP4 clip 841kB, hi-fi/original MP4 clip 2449kB.

Sources/Links

  • 2021-06-24: Does not compute: Avoiding pitfalls assessing the Internet's energy and carbon impacts: Although data traffic increased by more than 3-fold since 2016, network energy use dropped by 2.4% (Telefonica) and 50% (Cogent) over the same time period, demonstrating that historical network energy use is not directly proportional to data traffic. Notably, Telefonica's data traffic jumped by 45% in 2020 due in part to COVID (compared to 2019), with no reported increase in network energy use. Cogent’s electricity use dropped 21% from 2019 to 2020 even as data traffic increased 38%.
  • 2020-09-10: Video streaming: data transmission technology crucial for climate footprint: Fibre optic video transmission is nearly 50 times more efficient than UMTS. HD-quality video streaming produces different levels of greenhouse gas emissions depending on the transmission technology. The CO2 emissions generated by data processing in a data centre are relatively low, at 1.5 grams of CO2 per hour. ... The lowest CO2 emissions are produced when HD video is streamed at home over a fibre optic connection, with only two grams of CO2 per hour of video streaming for the data centre and data transmission. A copper cable (VDSL) generates four grams per hour. UMTS data transmission (3G), however, produces 90 grams of CO2 per hour. If the transmission technology used to transmit data is 5G instead, only about five grams of CO2 are emitted per hour. The electricity used by the end device is not factored into this calculation.
  • 2020-02-25: Factcheck: What is the carbon footprint of streaming video on Netflix?: Update 25/11/2020: ... the central IEA estimate for one hour of streaming video in 2019 is now 36gCO2 ....
  • 2017-01-31: The Internet Is Mostly Bots: More than half of web traffic comes from automated programs—many of them malicious.
  • 2015-11-26: What's the carbon footprint of an email?: 4g. "Receiving a spam message—even if you do not open it—has an environmental impact of 0.3 gCO2e." With ~10,000 attempted SPAM deliveries per day to my RPi implies ~3kgCO2e elsewhere in the network attempting to get those SPAMs to me...
  • 2012-01-24: The Cost of Spam: CO2 Emissions: suggests that in 2012 out of 220bn emails sent daily, 183bn were SPAM at an average carbon cost of 0.3g CO2 per SPAM for a total of 20Mt/CO2/year, compared to the entire UK in 2016 at about 381Mt.
  • 2009-04-16: Spam 'produces 17m tons of CO2'.
  • 2001-01-27: Press V for virus: "Like biological viruses, these [SPAM] messages need a host—or a cycle of hosts—in which to reproduce. In this case the hosts are us and our silicon boxes. And these viruses can cause a lot of disruptive and unpleasant side effects," including wasting time and energy, and thus spewing out carbon pollution...
  • Blocking robots on your web page – the list of 1800 bad bots.