9

I am concerned by people's privacy in the Internet age, and the security that communication software and services offer, which is why I try to promote encryption in different everyday activities (emails with PGP, communication apps that offer end-to-end encryption, encrypted data on hard drives...). I consider that those issues are part of how "sustainable" a society is (for reasons that might not need to be explained in details here).

However, encrypting and decrypting a file uses extra processing power, and therefore extra electricity, when compared to a regular file transmission or read-write activity. How important is that difference, and does it have a significant impact on the overall environmental cost of our general use of computers?

|improve this question
  • 1
    The answer would depend on many things, the length of time at 100% CPU, the make and model of the CPU, your current cost of electricity, etc – Some Free Mason Apr 7 '16 at 3:42
  • 2
    If you're using strong encryption, the environmental cost depends primarily on whether or not someone is expending the energy to try & crack it. – WBT Apr 8 '16 at 0:51
6

In fact there is an actual increase in energy consumption, though not much. What is done at the end devices is one thing (the encoding-decoding part). This answer from stackoverflow provides a good example:

When Google switched Gmail to use HTTPS, no additional resources were required; no network hardware, no new hosts. It only increased CPU load by about 1%.

But you also must take into account the protocol overhead (i.e. the additional information that must be exchanged between the 2 end points). Overhead means that more bits of information must be transmitted. Since communication networks are composed by several intermediate devices (e.g. wireless access points, switches, routers, .etc), there is an extra energy consumption, due to transmission and processing, at each of these nodes. Regarding the overhead added by a specific encryption protocol (TLS), some details are provided here. You may use this information to compute an average overhead (in %).

As a communications engineer, and without having made any detailed calculation, I tend to agree with the other answers, the additional environmental cost seems to be negligible.

|improve this answer
  • 1
    In the specific case of the devices I work with, we pay by the byte for GPRS transmissions, including UDP packet headers, so we count bytes a lot. Using AES128 gives a 16 byte block size, meaning in theory we "round up" or pad by an average of 8 bytes on a packet. But a packet has 28 bytes header, then we add 8 bytes of non-encrypted data plus a minimum of one encrypted block. So encryption adds 28+8+8+8 (the last 8 being the extra padding, or 8/44ths of the total ~20%. – Móż Apr 8 '16 at 3:41
  • 1
    In practice we bit-pack when we get close to the boundary, and have a bunch of encrypted data that goes in every packet, so the padding is only 2-3 bytes. With sloppier systems you get much the same benefit by zipping and optimising, which places like Google do intensively. SE itself has published data on this. People do it to save bandwidth, but encrypting less data saves energy, the question is whether it saves more than the cost of optimisation/compression (which is almost certainly beneficial for do-once stuff like image optimisation) – Móż Apr 8 '16 at 3:44
3

AFAIK there hasn't been much research on this topic, at least not about energy consumption of encryption on a global scale. Most research I've seen is concerned with the impact of specific types of encryption on the battery life of wireless devices (e.g. this article or this article)

Estimates are that the ICT sector is responsible for around 0.6% of the world's total energy use and about 4.7% of the generated electricity (see also this post). So on a global scale the impact of encryption is most likely rather small.

Suppose encryption is responsible for 0.01% of all ICT electricity usage (I'm not sure, just a wild guess on my part), that's 930 TWh * 0.0001 = 93 GWh or 93,000,000 kWh used per year.

Let's assume a carbon emissions factor of 0.5 kg / kWh (rough global estimate), then we can say that encryption is responsible for 93 * 1000 * 0.5 = 46500 metric tons of CO2e each year.

According to the EPA that's roughly the equivalent of 110,714,286 miles or 178,177,372 kilometers driven with an average American passenger vehicle, or the energy used by 4243 American homes in one year.

This is of course just a back-of-the-envelope calculation and depends largely on my estimate that encryption uses 0.01% of all ICT. If someone has a better estimate I'd be happy to revise my calculation.

|improve this answer
2

Not much at all. Since we waste more electricity waiting for a page to load with your monitor on. You would probably reduce your environmental cost by reading faster and not staring at the screen to think - leaving your monitor off as much as possible. This may help regarding the over all cost

Using a tablet to read and recharge the batteries with a portable solar powered charger is probably the most sustainable solution. Do you really NEED a desktop with an energy sucking harddrive?

Just a tip, offload your PGP encryption for emails by using an encrypted email provider. Hushmail is an example of that. While they WILL turn over your encrypted emails to comply with law enforcement, if you're just trying to stop snooping and are not breaking the law... you should be fine.

|improve this answer
  • How does the efficiency compare when you offload, though? If you're using HTTPS especially, the round-trip to an offsite decryption farm means that rather than "local PGP decryption" you do "HTTPS encrypt-send-HTTPS decrypt-PGP decrypt-HTTPS encrypt-send-HTTPS decrypt" which doesn't seem like a win. If you use a provider that receives and stores your email like Hushmail you leave out only the "HTTPS encrypt-send-HTTPS decrypt" steps, the rest is still there. So their PGP decryption would have to be extremely efficient to make it worth while. – Móż Apr 13 '16 at 1:25
  • You are over thinking this. There isn't much difference – Some Free Mason Apr 13 '16 at 1:35
  • "Not much difference" meaning it "uses only twice as much", or since only ~1% of the energy is used for encryption even doubling it doesn't matter? – Móż Apr 13 '16 at 3:10

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.