Cool experiment! Recently I’ve been more interested on energy consumption even if I don’t run devices with batteries.
However I think is misguided to think that getting energy from a “small” solar panel is the most environment friendly option. Most efficient is just to take energy from your personal power grid if possible! Reuses existing infrastructure which is already needed and big renewable plants are more efficient than small ones. And you don’t need batteries, since the grid can compensate itself (and if they have to store energy, they’re working on more efficient ways than a lithium battery). I don’t know in the US, but in my country it’s possible to have your home electricity at a company that only sells renewable energy, this if of course, virtual, since there are not different cables for green energy and not-green, but in the daily energy auction, it matters, your company only pays kW to green producers.
I don’t see OP making any implication that solar is the “most” environmental option, this piece seems to be more of a personal exploration and conversation starter that actually goes to some pains to not make big claims about either energy efficiency or environmental impacts.
Also I’m not sure an energy company stating that its sources are green / renewable is any more reliable than chocolate companies stating they use fair trade / ethically-sourced cocoa, when they’re both aggregating from mixed sources they won’t clear identify and that you can’t directly audit.
The electricity market is probably one of the hardest ones to cheat. Almost all production is national (and imports, when they happen, are from other EU countries), is centralized (the power grid administrators know each input where it comes from) and it’s being monitored in real time by several third parties. In fact, every person gets in their bills a graph with the sources of their energy (how much % was wind, how much % was nuclear, …) and this is mandatory. Nowadays, it’s also mandatory to have a digital electric meter at home, which are more precise with this stuff. In the end, as I said, everything is virtual as you can’t distinguish where it comes from but you can be sure that your money only goes to renewable producers.
Congratulations on living in a country where that’s (ostensibly) true — note that it’s always possible your faith in your government and its mandates is misplaced — most of the world’s population (including most living in the world’s largest energy economy) pays for power from for-profit companies with distinct and powerful incentives to lie, coupled with very little mandatory transparency and questionable at best regulatory enforcement.
I live in the UK, and post-Brexit and post-the-government-that-accidentally-Brexited I certainly wouldn’t take such a graph as reliable, much less gospel.
Here in the US, you can pretty easily pay for “net renewable” energy, where the power company pledges to buy at least as many kWh as you use from a renewable source. However, they time-shift this, buying solar and wind when it is cheap and available. When it is dark out, you are still getting your power from a non-renewable source.
I’m a little entertained that PHP and MySQL do run on the Raspberry Pi Zero 2W.
I found myself getting my back up about the personal websites not needing uptime part.
I personally favor an internet where I can go to someone’s personal website and expect it to be there. I’d rather not fall into an internet, which we have today, where only corporations can be expected to serve you content that you’re looking for. If I’m trying to reference a post for something I’m doing, it’s really helpful that the site I’m looking for is up. And I’d rather learn from other individuals than I would from the latest corporate ad supporting blog.
I found myself getting my back up about the personal websites not needing uptime part.
It is certainly something of a radical take, but also consider the consequences: if a website needs uptime, it basically turns running a personal website into a job. Suddenly you have users that you are responsible for. If you allow it to sometimes go offline, then it changes how you think about it. For example, if a site is offline, why is there not a recent mirror of it on something like archive.org? It’s fun sometimes to flip an assumption on its head and see what happens, instead of asking “how do I keep this on all the time?” you get to ask “what does the cost-benefit curve look like when I don’t keep this on all the time?” It’s a side-project, it should be fun!
I think the first website I saw that did something like this was https://solar.lowtechmagazine.com/about/the-solar-website, which goes much deeper into the paint than this one does about analyzing what consumes power and optimizing for low power use. Its uptime is not perfect but is not bad either, I wish they had more long-term numbers but that link cites about 98% uptime over a year. Of course it’s in Barcelona rather than Boston, which helps, but I also look forward to seeing how this particular solar website survives the winter~
if a website needs uptime, it basically turns running a personal website into a job.
For uptime alone? I think there’s a good part of the “work” spectrum before you cross the dreaded job border. Especially if you’re not going for umpteen nines.
But there’s something here that fixes some of the issues with this style of deployment and makes the work much easier: Community. I mean, sure, your SBC+Solar website doesn’t use too many resources. But yours and mine? Yours, mine, and Catherine’s? Whereas that one SBC+Solor installation could easily run all of ours. And we could take care of the maintenance if someone is on vacation, or just too busy. Web community gardens.
I’m not sure I really follow this reasoning. I have a personal website which is basically read only (no interactive features) but the services it runs require it to be 24/7. But as it’s hosted on a VPS it’s basically hassle-free for me since setup. Having to manage a solar charger, batteries etc. sounds like way more work, even if it’s only during the daylight hours.
For example, if a site is offline, why is there not a recent mirror of it on something like archive.org?
This is just having a online server 24/7 with extra steps.
Edit if this is what you want to do, go for it. Doing stuff for fun is what it’s all about, after all. But a server is literally designed to be running all the time. Having to deal with regular shutdowns is just adding complexity.
Oh, absolutely! And IMO a VPS is a really good solution, since it lets you get a lot of the benefits of economies of scale without needing to actually operate at that scale. I’m not rushing to shut down my own VPS, certainly. It’s just fun to see what the world looks like if you change some assumptions about how things need to operate. 100%-ish uptime takes a data center, UPS, backup generator, dedicated staff, etc. 98%-ish uptime takes a solar panel and battery that fit on a desk… plus a lot of infrastructure outside of that which is unaccounted for. Hmmm, makes you think…
Will that actually work acceptably? To me it seems like it does a surprisingly good job! Will that remain true through a Boston winter? Let’s find out!
Cool project and the solar panel dashboard in a variation of the GitHub activity overview is awesome!
When presenting the voltage over time, it actually does not reveal much information. The more current you draw from a panel, the more the voltage drops. If you draw peak current, your voltage approaches zero. On the other hand, with just a little bit of light, you can have a proper voltage but you won’t be able to draw any current.
I think an improvement would be to show either the current that is used to charge the battery, or the wattage.
The sad part is that most energy is consumed in the switching hardware between the server and client. There are huge shadownets spanning cities (with Wi-Fi and directional antennas) in Cuba, and even larger ad-hoc networks in Ukraine where even the switching is provided by the participants. This would be, in my mind, a much more consequential approach: Can you sustain a network, e.g. across a city, just with solar-powered components?
The sad part is that most energy is consumed in the switching hardware between the server and client
I wonder if this is true.
Based on eyeballing the allocation of space and power in homes and data centres, and the amount of kit needed for metropolitan-area nets, networking uses a lot less power than computing.
Some searchengineering suggests a typical cellular base station covers about 1000 users and uses about 1kW, and base stations are about 50% of the power usage of a cellular network, so that’s about 2W per user. Which is relatively large compared to the phone and the share of the server(s).
There was a nonprofit in my city called Metamesh that built a community wifi network like this, through a combination of grants and individual volunteers running their own nodes. It was pretty cool! Unfortunately it seems defunct now.
Meta Mesh founding president here, no longer on the board but still active where I can in promoting it. Pittmesh is essentially no more, as we couldn’t really find funding to maintain it or to build it out further. The non-profit organization pivoted to being a (Wi-Fi) internet service provider under the name Community Internet Solutions. it currently serves almost 150 customers but is going through some funding problems and could really use some donations while it’s awaiting some grant funding anticipated but not yet delivered.
Oh man, that’s too bad! I am always interested in mesh networking. There has been a resurgence of interest recently with LoRa-based Meshtastic, although that is much more bandwidth limited and will likely never serve websites. I’m building a few solar nodes this weekend to try to fill in coverage in our local mesh.
Super cool! One thing I’ve always wondered though in terms of energy usage is if its not more energy efficient for a small server to run in a VM, sharing a machine with other small servers, rather then on a dedicated machine that maybe is overprovisioned? Wouldn’t you get the most out of the hardware then?
What I do find sad is that on virtualized servers you can’t understand the electricity usage, as that information renders the servers on the machine insecure.
So cool. I always wanted to tinker with solar + efficient software choices but am worried about the space requirements. It looks like this space isn’t too different from mine. Maybe I should give it another look.
Cool experiment! Recently I’ve been more interested on energy consumption even if I don’t run devices with batteries.
However I think is misguided to think that getting energy from a “small” solar panel is the most environment friendly option. Most efficient is just to take energy from your personal power grid if possible! Reuses existing infrastructure which is already needed and big renewable plants are more efficient than small ones. And you don’t need batteries, since the grid can compensate itself (and if they have to store energy, they’re working on more efficient ways than a lithium battery). I don’t know in the US, but in my country it’s possible to have your home electricity at a company that only sells renewable energy, this if of course, virtual, since there are not different cables for green energy and not-green, but in the daily energy auction, it matters, your company only pays kW to green producers.
I don’t see OP making any implication that solar is the “most” environmental option, this piece seems to be more of a personal exploration and conversation starter that actually goes to some pains to not make big claims about either energy efficiency or environmental impacts.
Also I’m not sure an energy company stating that its sources are green / renewable is any more reliable than chocolate companies stating they use fair trade / ethically-sourced cocoa, when they’re both aggregating from mixed sources they won’t clear identify and that you can’t directly audit.
The electricity market is probably one of the hardest ones to cheat. Almost all production is national (and imports, when they happen, are from other EU countries), is centralized (the power grid administrators know each input where it comes from) and it’s being monitored in real time by several third parties. In fact, every person gets in their bills a graph with the sources of their energy (how much % was wind, how much % was nuclear, …) and this is mandatory. Nowadays, it’s also mandatory to have a digital electric meter at home, which are more precise with this stuff. In the end, as I said, everything is virtual as you can’t distinguish where it comes from but you can be sure that your money only goes to renewable producers.
Congratulations on living in a country where that’s (ostensibly) true — note that it’s always possible your faith in your government and its mandates is misplaced — most of the world’s population (including most living in the world’s largest energy economy) pays for power from for-profit companies with distinct and powerful incentives to lie, coupled with very little mandatory transparency and questionable at best regulatory enforcement.
I live in the UK, and post-Brexit and post-the-government-that-accidentally-Brexited I certainly wouldn’t take such a graph as reliable, much less gospel.
Here in the US, you can pretty easily pay for “net renewable” energy, where the power company pledges to buy at least as many kWh as you use from a renewable source. However, they time-shift this, buying solar and wind when it is cheap and available. When it is dark out, you are still getting your power from a non-renewable source.
super cool!
start a solar blogring with the solar protocol folks :)
https://solarprotocol.net/
I’m a little entertained that PHP and MySQL do run on the Raspberry Pi Zero 2W.
I found myself getting my back up about the personal websites not needing uptime part.
I personally favor an internet where I can go to someone’s personal website and expect it to be there. I’d rather not fall into an internet, which we have today, where only corporations can be expected to serve you content that you’re looking for. If I’m trying to reference a post for something I’m doing, it’s really helpful that the site I’m looking for is up. And I’d rather learn from other individuals than I would from the latest corporate ad supporting blog.
It is certainly something of a radical take, but also consider the consequences: if a website needs uptime, it basically turns running a personal website into a job. Suddenly you have users that you are responsible for. If you allow it to sometimes go offline, then it changes how you think about it. For example, if a site is offline, why is there not a recent mirror of it on something like archive.org? It’s fun sometimes to flip an assumption on its head and see what happens, instead of asking “how do I keep this on all the time?” you get to ask “what does the cost-benefit curve look like when I don’t keep this on all the time?” It’s a side-project, it should be fun!
I think the first website I saw that did something like this was https://solar.lowtechmagazine.com/about/the-solar-website, which goes much deeper into the paint than this one does about analyzing what consumes power and optimizing for low power use. Its uptime is not perfect but is not bad either, I wish they had more long-term numbers but that link cites about 98% uptime over a year. Of course it’s in Barcelona rather than Boston, which helps, but I also look forward to seeing how this particular solar website survives the winter~
For uptime alone? I think there’s a good part of the “work” spectrum before you cross the dreaded job border. Especially if you’re not going for umpteen nines.
But there’s something here that fixes some of the issues with this style of deployment and makes the work much easier: Community. I mean, sure, your SBC+Solar website doesn’t use too many resources. But yours and mine? Yours, mine, and Catherine’s? Whereas that one SBC+Solor installation could easily run all of ours. And we could take care of the maintenance if someone is on vacation, or just too busy. Web community gardens.
I’m not sure I really follow this reasoning. I have a personal website which is basically read only (no interactive features) but the services it runs require it to be 24/7. But as it’s hosted on a VPS it’s basically hassle-free for me since setup. Having to manage a solar charger, batteries etc. sounds like way more work, even if it’s only during the daylight hours.
This is just having a online server 24/7 with extra steps.
Edit if this is what you want to do, go for it. Doing stuff for fun is what it’s all about, after all. But a server is literally designed to be running all the time. Having to deal with regular shutdowns is just adding complexity.
Oh, absolutely! And IMO a VPS is a really good solution, since it lets you get a lot of the benefits of economies of scale without needing to actually operate at that scale. I’m not rushing to shut down my own VPS, certainly. It’s just fun to see what the world looks like if you change some assumptions about how things need to operate. 100%-ish uptime takes a data center, UPS, backup generator, dedicated staff, etc. 98%-ish uptime takes a solar panel and battery that fit on a desk… plus a lot of infrastructure outside of that which is unaccounted for. Hmmm, makes you think…
Will that actually work acceptably? To me it seems like it does a surprisingly good job! Will that remain true through a Boston winter? Let’s find out!
Judging by the battery levels, it could also power a bigger server. I wonder whether this is also the case in winter. We’ll probably see …
We’ll see indeed!
Cool project and the solar panel dashboard in a variation of the GitHub activity overview is awesome!
When presenting the voltage over time, it actually does not reveal much information. The more current you draw from a panel, the more the voltage drops. If you draw peak current, your voltage approaches zero. On the other hand, with just a little bit of light, you can have a proper voltage but you won’t be able to draw any current.
I think an improvement would be to show either the current that is used to charge the battery, or the wattage.
The sad part is that most energy is consumed in the switching hardware between the server and client. There are huge shadownets spanning cities (with Wi-Fi and directional antennas) in Cuba, and even larger ad-hoc networks in Ukraine where even the switching is provided by the participants. This would be, in my mind, a much more consequential approach: Can you sustain a network, e.g. across a city, just with solar-powered components?
I wonder if this is true.
Based on eyeballing the allocation of space and power in homes and data centres, and the amount of kit needed for metropolitan-area nets, networking uses a lot less power than computing.
Some searchengineering suggests a typical cellular base station covers about 1000 users and uses about 1kW, and base stations are about 50% of the power usage of a cellular network, so that’s about 2W per user. Which is relatively large compared to the phone and the share of the server(s).
So, hmm, dunno.
There was a nonprofit in my city called Metamesh that built a community wifi network like this, through a combination of grants and individual volunteers running their own nodes. It was pretty cool! Unfortunately it seems defunct now.
Meta Mesh founding president here, no longer on the board but still active where I can in promoting it. Pittmesh is essentially no more, as we couldn’t really find funding to maintain it or to build it out further. The non-profit organization pivoted to being a (Wi-Fi) internet service provider under the name Community Internet Solutions. it currently serves almost 150 customers but is going through some funding problems and could really use some donations while it’s awaiting some grant funding anticipated but not yet delivered.
Oh man, that’s too bad! I am always interested in mesh networking. There has been a resurgence of interest recently with LoRa-based Meshtastic, although that is much more bandwidth limited and will likely never serve websites. I’m building a few solar nodes this weekend to try to fill in coverage in our local mesh.
Super cool! One thing I’ve always wondered though in terms of energy usage is if its not more energy efficient for a small server to run in a VM, sharing a machine with other small servers, rather then on a dedicated machine that maybe is overprovisioned? Wouldn’t you get the most out of the hardware then?
What I do find sad is that on virtualized servers you can’t understand the electricity usage, as that information renders the servers on the machine insecure.
So cool. I always wanted to tinker with solar + efficient software choices but am worried about the space requirements. It looks like this space isn’t too different from mine. Maybe I should give it another look.
I know I don’t have much to add, but what I do have - is important to say: This post is super inspiring, thank you!