Can Safe Network become the world's most energy-efficient Internet network? (Energy efficiency of the Safe Network)

Since the recent changes in the SafeNet project are so large that they will most likely significantly reduce the requirements for energy consumption, it is worth starting a discussion about the energy efficiency of SN.

Important - The definition of performance and efficiency.

First of all, it should be noted that the issue is efficiency, not energy performance, and that is the fundamental difference - performance is how effectively a system can achieve its goal, while efficiency is how much resources are used to achieve that goal.
To determine energy efficiency, you need first understand what affects them - It can be assumed, that there will be three main elements:

1. How much energy will SafeNet use for its operating activities?

Operational activities of Safe Net means:

  • maintaining data storage and transfer resources (hard drive capacity and bandwidth consumption),
  • maintaining the necessary amount of data copies (data replication and deduplication),
  • serverless network of distributed nodes operating on the basis of network consensus, based on a very efficient Proof-of-Resource mechanism (CPU power and sufficient bandwidth),
  • a payment system based on Digital Bearer Certificates (DBC), ensuring fast, direct payments without the intermediation of a central mint (including network fees using Safe Network Token),
  • automated, secure real-time software updates (with self-checking system to protect against malicious updates).

The estimation of the energy consumption necessary for SN operational activity, will be a basic component of the efficiency of the entire network.

2. Will SafeNet work as efficiently as ant colonies?

From the beginning, @Dirvine and the team intended to use the best solutions created by nature in the project, following the example of the most efficient creatures and organization, which are ants and ant colonies.

Currently, scientists agree in their studies that, apart from numerous unique features of ants, the efficiency of anthills is the result of… inactivity of the vast majority of ants :slight_smile: according to the less is more principle or the Pareto principle.
Depending on the size of the colony, this number is from 60 to 80% of workers and is mainly related to the limited capacity of the tunnels, but also to the rotational organization of the teams’ work.

If it is possible to determine to what extent the SN could map the functioning of anthills, it will be possible to estimate the energy consumption related to the transfer of portions and data backups.

3. How much power consumption does network security require?

An integral element of SN is privacy and security, which by default includes:

  • self-authentication of users (authentication data is never stored on the network),
  • hiding the client’s IP address and encrypted communication throughout the network (from the first connection to the SN),
  • division of files into parts, sending them to different locations and self-encryption (only having a data map allows you to read the file),
  • multi-level encryption (used, for example, in direct messages or when creating a public profile),
    and in addition, several other functionalities that overlap each other, increasing the level of network security.

Estimating the energy consumption necessary to achieve the main goal of the project (absolute privacy and data security) will show us the energy consumption of the security system.
However, this does not mean that this is the energy cost of security, because SN will inherently avoid many hacker attacks, and this also means saving energy that would be used to try to break security.
Examples of attacks that will be practically impossible or very difficult include:

  • Sybil attack - distribution of nodes, encryption, hidden location and migration of portions as well as detection of malicious nodes makes the attack very expensive in terms of energy (and not only), and the larger the network, the more expensive,
  • DDoS - an attack of many computers on the network would prove ineffective because there would be no central point of attack, and the cost of energy would be very significant,
  • Man in the Middle - an attempt to forge the browser’s certificate will be impossible because the Safe browser is built into the system,
  • Malware - a group of various programs that, similarly to a Brute Force attack, aims to identify data access data, damage system functionality, delete data, open the door to further attacks or block the computer, will have a very difficult task and will consume a lot of energy trying defeat security.

Of course, there are many more elements affecting the energy efficiency of SafeNet, such as energy consumption by various software languages (once @happybeing posted very interesting links to information on energy efficiency and I hope that he will bring them to the discussion).

So please correct me where I am wrong and I am asking for your ideas for estimating the energy consumption of the SN.


For me the best steps towards lowering energy consumption by Safenet lately is reduction of resources (CPU, bandwidth, memory) consumption, and dumping node ageing. So nodes from home are possible, and even better, nodes on mobile or IoT. So there is no need for dedicated large scale nodes, and everyone can start farming on their device using spare resources. I see an analogy here to an 80% idle ants, where many private nodes will work just part-time.

To assess the consumption it’s most important to ask yourself at what point we will measure it. Do we want to know the consumption at the beginning, or in 5 years from launch? The network will grow and energy consumption will grow as well! Or perhaps we could assume current Ethereum or Bitcoin scale / functionality and compare to that?


The power usage is an interesting one with many factors involved.

Some of the factors when considering moving from cloud storage to Safe Node storage.

Consider the data centre cloud storage where the providers are extremely conscious of their power usage since even a 1.0% overall reduction in energy used to power the computers has a large effect on their profit. For the home user though a 10% variance in their computer usage has little bearing on their overall power bill.

A Safe Node in certain cases will change the power profile of the home computer. The computer may be used for 4 or 8 or 12 or … hours per day for most days in a week and become a 24/7 always on so that a Safe Node(s) can run all the time. Initially most new home Safe Nodes will be run on a person’s computer and in some cases later on a SBC will be bought to run Safe Nodes on.

Profile of data access in cloud storage is speed and availability of access to the data. Profile of the Safe Network is essentially the same. The difference is in how this is achieved and what is considered acceptable. The cloud storage relies on large internal networks connecting to large internet backbone connections with ultra secure links. Secure as in uptime and connectivity. Safe achieves this with redundant storage spread across the world and the world’s internet providers (thus backbone access) allowing cheap and less secure (uptime etc) links to be used. Per element of data retrieved the power required to provide this networking specifically for cloud storage and speed of operation is much higher than the power for home user.

Cloud storage provides differing levels of backup. Starts with the inherent reliability of RAID or better (eg Google email) all the way to multi site replicated storage w/backups and further. Safe provides multi site replication, RAID style storage. Archive nodes will provide similar levels of security effectively being off-site backup. This means the cloud storage providers are expending more energy in proving security of data than just storing the data. Whereas this is just part of the Safe Network.

Cloud providers also require large cooling infrastructure to keep the data centre operational. The Safe Node typically requires no more cooling than the home provider already is doing while computer is running. The Safe Node though will require the home computer to run 24/7 which might mean longer run times for the cooling (if needed). SBC nodes will typically require no additional cooling since the home owner lives there too and the SBC is a very low power device that its fan handles (if needed)

There are more but this is starters.

Now to some advantages of Safe Network not realised across all the cloud providers.

Data duplication With the advent of cloud providers increasingly providing encryption by the client and business’ machines we are seeing less opportunity for the cloud provider to gain benefit from even attempting to detect duplicate files. And never across cloud providers. Whereas Safe for the bulk of it storage (immutable chunks) will start to reduce the requirements for storage media for existing data. All those cat videos/pics will now be stored once to the network with the 4 or 8 replication. That one cat vid could be uploaded to FB 1000 times in a day and to twitter more times. That copy of a movie stored on 100,000 computers around the world will now only have 4 or 8 copies as people upload their film libraries to Safe. These are just illustrating the savings possible on certain types of data. All this saves energy in the long run.

Now depending on development and final requirements to run Safe Node, I can foresee new businesses taking a SBC (RPi, Odroid, Odyssey, etc) and configuring the SBC to run a Safe Node, adding a storage device and selling them to consumers. Maybe for the more powerful SBCs add in streaming services etc for the home TV. These draw minimal power in terms of less than 20 watts. And once this is occurring there will be a large expansion of people running a Safe Node due to the low total cost of ownership. These will be a viable alternative for those people not wanting to leave their home computer on for 24 hours a day. Thus a reduction across the network of the average energy per node.

In AU our electricity retail price is rising over 30% on the 1st of July and will be over 45 cents per KWH.

  • Home computer: (not typically on all day)
    • left on say an extra 10 hours a day at 200 watts is 2.0KWH
    • @ 45 cents/KWH is $0.90 per day
    • and $328.50 per year.
  • SBC:
    • average 12 watts (should be lower) is 0.288 KWH a day
    • @$0.45/KWH is $0.1296 per day
    • and $47.30 per year.
  • Difference:
    • $328.50 - $47.30 = $281.20
    • Assuming home PC is used daily normally. The savings are even more if home computer needs to on for longer hours than normal.
    • More than the basic price for many lower end SBCs with Drive
      • in the Odyssey’s case which is a high end SBC a media centre could be installed or used for browsing the web thus reducing the need to use the home PC every day.

Some ideas I had


We already know, that the larger the network, the lower the energy consumption per user or, for example, 1 TB of stored data, because it results from the nature of SafeNet’s construction and operation. So for now we can look for differences between the power consumption:

  • in the currently used solutions, and the solutions achieved so far by SN.

I deliberately did not mention the comparison of SN to ETH, BTC or even to the blockchain, because these solutions have many disadvantages and very high energy consumption, which disqualifies them as a global distributed Internet infrastructure available to everyone.
I think, that first of all, energy efficiency (and not only) should be compared to the current Clearnet, mainly due to the ease of comparison in a wide range.


Great post!
Many key elements have been explained that I didn’t mention in the first post because it would be discouragingly long, thank you for doing it for me :smiling_face:

Quite a long time ago I asked if computers (nodes) will have to be online all the time for the network to work, @dirvine explained, that they do not have to, and the entropy of the network is organized in such a way, that there is always some copy of the data (those that can be stored on turned off computer).
Have you made the assumption that the computer will be able to run several hours a day or will it have to run 24/7 to keep the node active?


Smaller nodes make this part easier for computers to be “part time” and all going well it’s a good goal to have those in poorer places with slow computers and dodgy power being able to particiapte and earn.


My laptop has 90W power supply, and my netbook – 30W. And that’s max power, when the device is charging. When charged, the power drops significantly. My laptop charges in 2 hours and discharges in 8, so I think we safely can assume 1/4 of max power supply, 23W, is drawn when discharging. And it’s with the screen on, so when screen is off, the laptop needs even less power.


Yes laptops are the inbetween. Desktops used for gaming or just with a decent sized GPU is what i was thinking of and in the first stage of adoption will make up a large percentage of desktops used for nodes. Later on the desktops that a writer or casual user will use won’t have the large average power draw.

It is difficult to cover all cases in a brief examination of some issues to look at. And as you point out laptops are also another case.

I reckon if needed I could end up with a 100 to 200 page brief report examining the ins and outs of computer usage and a comparison with current large scale storage. Power usage is quite varied just for the home computer situation.


I’m glad we’re having a think about this whole subject.

People think that Cloud providers and Datacentres in general are efficient and in many ways they are because of the consolidation of resources. But the big thing wrong with most of them is that they don’t use the waste heat they generate. There are a few that use it to heat houses, swimming pools, grow tomatoes, etc. but they are few and far between.

The other thing wrong with them is that most of them also use a significant amount of energy in moving that heat to the air outside. There are more modern designs that cool a small amount of air by running the chillers at their optimum speed and mixing that with air from outside during the cooler months of the year but the problem still remains that they are not using the waste heat and using more energy to move it outside.

(One of the maddest things I’ve ever seen was in a new Datahall in a large Datacentre. I was surprised to see a bunch of 5KW space heaters running in it! “Why?!” I hear you ask? Because otherwise the air cooling and conditioning system wouldn’t work properly! The Datahall was new and largely empty with just a couple of racks installed in the space that would eventually have hundreds of racks (we were installing some of the first racks in this big empty space). They needed to have the air conditioning and cooling system running to regulate the temperature and humidity across the Datahall for the servers that were running but it doesn’t function well when the air isn’t hot enough and furthermore when all the heat being generated is all up one end of the space. Seems mad but it makes perfect sense when you are already doing something definitely not optimal.)

I think that running Safe nodes at home has a big advantage over running compute and storage in datacentres that we’ve not talked about yet - the waste heat might not really be waste for most of the year. Here in the UK it’s only really between 3 or maybe 5 months of the year when most people don’t have the central heating on at least for part of the day depending on which end of this sceptered isle you live, energy efficiency of your house, age, health, personal preference, etc. The rest of the time something like 20W from a couple of Raspberry Pi’s with a couple of SSDs or even 200W from a desktop as a background level of heat throughout the day could be considered a benefit.

When my gas boiler has had issues in the dead of winter I have hibernated in the living room with multiple computers, XBox, Playstation, TV, Hi-Fi all on and it makes a big difference.

I’m wondering though if we’ll find that storage capacity flows around the world with the seasons to some extent?

Obviously there are places where more heat in the house is never welcome and they tend to be places where solar power works well. So that’s a conundrum. Maybe their Safenodes won’t even be in the house?


This is a great example of how you can reduce the power consumption of a device that will run Safe nodes, while extending their operating hours to 24 hours a day.

Update (once the assumptions about how to use the computer have been clarified).

The difference in power consumption will be as follows:

Home computer: (people who have higher-end PC computers, gamers, coders, technology enthusiasts and need to have GPUs like 2080, 3080, 4080 - used for 14 hours for other reasons, including keeping the Safe node with negligibly low power consumption)

  • left on for, say 10 hours a day is extra computer work just to maintain a Safe node at 200 watts is 2.0 kWh
  • 2.0 kWh x 365 days = ~ 739 kWh per year.
    SBC: (assuming it runs 24 hours a day)
  • An average of 12 watts (should be less) is 0.288 kWh per day
  • 0.288 kWh x 365 days = ~ 105.12 kWh per year.
    PC (in use extra 3650 hours per year) vs. SBC (in use 8760 hours per year):
    739 kWh - 105.12 kWh = - 633.88 kWh !

Keep in mind that you are an advanced user, rather at the vanguard of the Internet :wink:
The average user, and there will be (I hope) hundreds of millions of them, have museum computers, which are often very neglected and therefore consume more power, and the power consumption depends on the activities performed.
So, in my opinion, the assumed power consumption of 200 Wh for advanced computers and power consumption of ~150 Wh for non-technical people’s computers s justified.


I actually was looking at the home PC still being used as it was with the SBC running Safe separately. So my maths was for examining the difference in ON time for the PC rather than comparing running a PC solely for Safe compared to SBC. Basically showing that buying a SBC might be a viable option rather than running the PC for the extra hours.

But as was pointed out the Safe network design now allows for nodes to run part of day only and kind of makes my example only apply for those wanting to run the nodes 24/7

This points out another area where the desired outcome will require different analysis on the power difference and how trying to explain things briefly doesn’t always bring clarity. Your example seems to compare the PC sole reason for being on is for Safe node running. My example the PC was on in any case for other reasons and looking at extending the on time.

Both are good examinations.


I see, I just assumed that your calculation of the cost of energy consumed was based on analogous use (similar operations) of PC and SBC.

In this case, what values should we use to calculate the difference in energy consumption / cost in a reliable way?
(To sow Safe farming and not mythomania) :wink:

1 Like

I thought using the regular use of the computer vs the regular use + node.

Then compare the difference between the 2 figures against other solutions/implementations

My reasoning is that is what people will be concerned with when working out if its worth running a node. Also for working out the difference between cloud and safe on home computers

1 Like

Sorry to bother you with this as you probably have more important things to do, but I would like to reliably calculate this difference in power consumption so that other users can understand.

I understand that you mean that the PC is used for 14 hours for other reasons, including maintaining a Safe node, whereas 10 hours a day is the extra work of the PC just for the purpose of maintaining a Safe node - i.e. both PC / SBC devices are working 24 hours a day - have I got this right?

And if so, is the assumption of a PC power consumption of 200 watts just to maintain the node - not too high?

And if the above is correct in terms of energy consumption (as well as cost), it should be:

  • Home computer - used for normal applications (including node maintenance) 14 h + additional time just for node maintenance 10 h.
  • SBC - used for node maintenance only 24 hours.

Let’s flip it around…

What if Safe was the most effienct Energy Network?

(I’ll just leave that there :wink:)


This is the answer I have been waiting for… (exactly from you) :wink:

Because that’s one of my goals for what I expect from SafeNet.


Yes. Normally the PC is off for 10 hours on a typical day which is the scenario for this example and i expect this is at the high end for those who work/school and turn off their computers when at work/school and sleep.

I chose this as the higher end for this scenario. My thoughts were that the people in the early days who are most likely to run nodes will be those who also have higher end PCs, maybe gamers, maybe coders, maybe love technology and must have 2080s, 3080s, 4080s type GPUs. Some of these though leave their PCs turned on all of the time and that analysis doesn’t include those, except they may change their ways and get a SBC to save some money. I have requested my adult son turn off his PC while he sleeps to reduce our electricity bill.

Depends on the information you wish to show.

for what you say that is total consumption comparison and mine was additional power consumption. Mine was to consider if a SBC was potentially worth the cost to buy and setup. For that i needed to compare the additional cost required for the node. I did assume that during the 14 hours the node operation was minimal additional cost on a higher end PC as I would expect anyhow.

Now for a PC that maybe a non gamer, and non tech person has the power might be 150 Watts maximum while using the PC (w/screen) and 80 or 90 watts while on but in idle state (screen off after 5 minutes etc). And running a node on that machine might increase power usage by 5 watts average (now significant) and total power used would perhaps be a better comparison. On the order of 4 to 5 percent whereas for the high end PC with multiple drives it might not even be 1 watt extra and in the “noise” level of measurement.

But if one was to do a rigorous examination then researching studies done on home usage of PCs would be needed to have say 5 representative power/usage profiles of home setups and then 3 or 4 representative on/off times for the PCs. That would end up with a table showing the 15 or 20 profiles etc.

In the end if Safe achieves the performance expected (we hope for better LOL) then I’d expect over time there will be a reduction in growth (maybe even negative growth eventually) or repurposing of a significant percentage of data centres, especially the more inefficient ones. Will not be overnight and we also must remember that they are still building datacentres at a high rate at the moment and the momentum will continue during the early stages of the Safe Network.

Basically there will be two growth curves, one for datacentres holding cloud storage systems and one for the Safe network. The growth of datacentres will reduce as Safe Network grows past the slow adoption phase. And by then we will start seeing a drop in the energy required to store each MB (or PB if you want) on a global scale.

I doubt data centres will disappear as decentralised data storage of any kind does not fit some applications. The NSA for instance and our (maybe many/most) Government(s) will not allow their data to be stored outside of the(ir) country.


Energy-efficiency might be less of an issue as well when we start to use more and more renewable sources. This is a whole discussion on it own of course but I see Safe as a long term technology and so maybe we should look at energy with a long term perspective as well. Price might drop drastically.

However, prices for harddrives, electronic components might raise a lot when we start to incorporate fair pricing. Where the proper impact on the earth is calculated and taken into account with final pricing.

Thank you for the clarification and detail, of course this is not an exact analysis but a fair calculation of energy consumption and savings - I will update the description and complete the calculation with a PC option for less advanced users.

I am not 100% sure, but I think @JimCollinson had something else in mind :wink:

But I totally agree with what you wrote.

1 Like

Off-topic… whatever happened to that private technology that was going to use lots of small devices to transmit power to each other on the move? Around six years ago, there was some hype, anyone remember it?