Somewhat tangential, but in my first NYC apartment, our radiator was absolutely terrible, and did a miserable job getting the living room to an even remotely livable temperature.
My solution? Two giant rack mount servers, running Foldign@Home at full capacity. Way I figured is that it's still 100% efficient at generating heat, but I might as well help drug research in the process.
that's the opposite problem that most people (from my experience) have in NYC. Apartments get too much heat (i.e. landlords can get seriously fined if they dont keep the building at a proper temp, so its just easier and safer for them to give too much heat). Also, many radiators in old buildings aren't really adjustable anymore (and hence you get all the heat). so its common for people in NYC to keep their windows open in the middle of winter to try to balance out the temperature.
We don’t live there anymore, so we don’t, but we honestly just put up with it. I don’t have any pointers on that fact.
To the child comment; I’ve told this story in the past but the tldr of it was that I was (unknowingly) not paying for the electricity in my apartment due to a broken meter. I didn’t break it, and I didn’t know it was broken; I just thought power was really cheap in NYC.
I'm highly skeptical this thing can provide enough heat energy to warm an entire water tank and keep it warm for the sporadic on-demand loads a hot water tank typically endures. Will it take 3 days to warm back up after a moderately long hot shower while simultaneously running the dishwasher?
I'm further skeptical at the claim of a dedicated fiber line being installed just for this server (having been on the receiving side of fiber install quotes with 5-6 digits) - and even more skeptical at the claim of using 4G/5G connections for servers.
I'll believe it when I see it... but it reads like it was written by someone that really doesn't have a grasp of what we're dealing with here.
I clicked through to the trial and in the FAQ it says:
Will you use my broadband?
Yes, we will connect to your broadband to communicate with the units.
We'll directly connect to your router to avoid WiFi interference and most of the time the unit will simply be sending some monitoring information (temperatures / fan speeds etc) back to base, so you shouldn't notice any impact.
We'll occasionally perform speed tests, and limit any larger uploads and downloads to a fraction of this so we'll never use too much of your bandwidth. We'll also schedule these overnight so it shouldn't affect your daytime speeds.
Our intention is that you won't notice - one of the things we'll be doing during the trial is getting your feedback on if we've achieved this.
In the future we plan to install a dedicated fibre or 4G/5G connection for the heata unit.
How useful is a compute node that can't use the internet? I suspect the entire truth is not being told here - not to mention this is in direct contradiction to what the article author claimed.
And if it does use the internet, it's now using your internet (and presumably your electricity) for free. Seems heavily lopsided in the company's favor, geared towards selling suckers on feeling good about "helping" the environment when you are really just helping some company make money.
> How useful is a compute node that can't use the internet
It doesn't say that. It says that it will use your broadband connection. It says most of the time it will just be for monitoring and speed tests, and it also says that larger uploads/downloads will be limited to a fraction of your bandwidth and they'll try to run it overnight.
To me, this says they'll send/retrieve datasets to be processed on the server, slowly, mostly overnight.
> It is in direct contradiction to what the author of the article wrote
Yes. I don't know why he wrote it, but if you click through to the trial on https://www.heata.co/trial and scroll down to the FAQ, it is right there.
> And if it does use the internet, it's now using your internet (and presumably your electricity) for free
I think maybe you need to read the FAQ. They see how much electricity they're using by installing a meter for this specific purpose, and they pay you for it.
> It doesn't say that. It says that it will use your broadband connection. It says most of the time it will just be for monitoring and speed tests, and it also says that larger uploads/downloads will be limited to a fraction of your bandwidth and they'll try to run it overnight.
This doesn't seem very useful at all. Other than to get feel-good enviro points, and even then it's still highly questionable.
If it can only use a small fraction of your bandwidth, and it can only do it at night - then who is this thing for? What kind of profitable compute keeps it occupied all day and only checks in at night?
I highly suspect the entire truth is not being told here with this idea. It sounds like a half-baked college student idea more than a professional operation.
Some jobs require a large amount of cpu and a small amount of data. I imagine that is what this system is for.
Ultimately, it will be costing them money to create this hardware, install it in your house, and then pay you for the electricity it uses. It seems likely to me that they'll have considered a way of making money from it.
[edit] I investigated further and one of the things they do on these machines is 3D rendering.
Wouldn't this destroy my upload speed? Isn't running a regular server (not home media server) rate limited/outright banned in the TOS of most home internet service providers?
They specifically say that they run speed tests and then restrict their usage to a fraction of your speed. So no.
> Isn't running a regular server (not home media server) rate limited/outright banned in the TOS of most home internet service providers?
Don't think of it as a server. Think of it as a PC that is running all the time. These TOS you're thinking of are usually to prevent you running services that can be connected to like web servers and smtp servers. For all we know, this machine doesn't listen for incoming connections and just makes outgoing connections. Just like your phones browser is doing. Your phone isn't a server banned by your ISP's TOS is it?
You may want to re-read; this isn't disabling the current water heater, it is modifying it to be less insulated, and placing a compute unit there. [It will maintain an immersion heater]
Re: fiber, Natural gas lines often already have a fiber line for monitoring (some US utils were the first to the fiber market because of this)
The cost of routing a fiber line is extraordinary. Even if one existed nearby, it would not be usable because it's owned by the utility not this private server company.
Regarding keeping the existing heating element - seems necessary because this server water heater isn't going to heat much of anything.
To flip it on it's head - this is free cooling capacity for a for-profit server company that will do little-to-zero for the home owner and risks damaging their expensive water heater for some hand-wavy promised savings of $200 a year... not really selling me.
They... would just get a contract with the utility ?
Also, I don't think you just get one fiber(?), I have 4 fiber ports on my wall outlet, which was quite useful when I changed ISPs (that didn't share the same fiber) : only a few seconds of downtime, to switch Ethernet cables and Wi-Fi !
Also, unless you have solar thermal hot water, this sounds to be the perfect use case, thanks to the large inertia... (I wonder how they interface with the off-hours cheap electricity? This seems to be up to a constant 200W, ~100W guaranteed ?)
The concern I have here is that despite the "patented heating element", this sounds to be a hacky solution that might compromise the insulation too much. They really need to build those in the water heater to start with ! (And provide regular service against failure from rust, since a leaky heater might have more dire consequences here.)
There is no reality where this server "heater" can replace a natural gas or electricity line - so the comparison is moot.
Further, fiber lines are owned by an entity. This private, for-profit company is not going to be able to just use someone else's fiber lines simply because they are nearby.
Even further, bringing a fiber line even across a building can run thousands or tens of thousands. Across a street? Try a hundred thousand... I've been here many times with fiber, ie. there's nothing cheap about it, all of it.
There is no reality where this server company is going to shell out tens of thousands to run fiber lines to a bunch of one-off installations.
Ah, you've missed the context that this is coordinated with British Gas, providing the utility natural gas infrastructure that exists to heat the water.
Many gas companies have already entrenched fiber when performing maintenance on consumer gas lines for monitoring stations. Fiber is shallower and is easy to trench relative to gas lines. It also serves as an early warning sign when digging.
None of which means this for-profit company gets access to utility infrastructure. Even if they did... it would still be cost prohibitive to split, splice, run and terminate fiber into a residential unit - all in the name of saving $200 a year.
It simply does not make economic sense. Like I said before - I strongly suspect the entire truth is not being told about this arrangement.
From the FAQ on the trial page, it seems like it will simply not run if there is nowhere to dump the excess heat:
How much of my hot water will it deliver?
A useful base load. If the heata unit ran all day it would deliver about 4.8kWh of hot water which is about the amount of hot water an average household uses. In practice the utilisation level will be lower and sometimes the water will be at temperature so the heata unit won't be able to run. We think it could deliver up to 80% of an average household's hot water, with good compute utilisation. This trial will help us learn more about how this will work in practice and enable us to give you a useful amount of free hot water at the same time.
Sounds like it is connected to an existing hot water tank with the conventional heating mechanism left intact (I could be misreading, but it sounds like that given a hole is cut into the heater to make room for the additional heat sink)
I believe they mean dedicated fibre as in a home fibre broadband connection. In the UK that can mean anything from VDSL to FTTH (thanks advertising watchdog!).
These aren't being used as web servers and the like. The intention is to take batch jobs and run them - computer modelling etc. The data can easily be downloaded overnight over even a relatively modest connection.
Worth remembering that you don't need the server to completely heat the water to shower temperature. You can get it part way there, and then it will take less heat to get the water heater the rest of the way.
I could imagine a future where every house has something like this partially heating their water and air before the real heating system is applied.
> I could imagine a future where every house has something like this partially heating their water and air before the real heating system is applied.
This is really important.
Where I live - Oslo, Norway - we have a sun thermal collector-thingy[0] on the roof which _aids_ in heating up other stuff. Surely, it needs to be a fire nearby to heat the water enough to have a comfortable shower, but we save approx. $20K yearly[1] on energy costs with this solution.
[1] I live in a housing association with 60 other families in the same building, so needless to say the savings would be a lot less if I lived on my own in one house. :)
This could be flipped - I.e. using a Japanese domestic fuel cell[1] one could install a hot water heater in the home, and get 'free' electricity to run a server, that could possibly do something productive to offset the associated capital and operational expenditures.
(Interesting aside: some people in Japan combine these with a solar/battery array to achieve robust all-season off grid self sufficiency, without combustion, which I think is quite neat!)
True, though it'd be using propane / LP gas, where the server-cooled-by-heating-your-water could potentially be powered by electricity from solar or wind...
Doing some back of the napkin math, their contractual minimum of 2.5 kWh a day into your water heater works out to 75 kWh a month. This [1] site claims the average two person home uses 195 kWh a month, so it potentially could be a decent offset. I do wonder whether this quoted 2.5 kWh is energy coming off the server into your heater, or if it's measured somewhere else. Energy loss from thermal transfer would make a huge difference in measured input into your heating system.
very clever but I wonder what kind of workloads are being run on these remote/edge compute nodes. there isn't any homogeneity in the network - everyone is going to be on a different kind of network with a huge variations in bandwidth and latency.
can't find any specs on what kind of compute power each physical installation would provide
The compute rig would need to be the size of the water tank to make the business even.
Maintenance cost would be much higher too. Better to keep all compute in house sized DC units and just sell hot water to neighbors.
These guys [1] used to be called Nerdalize and iirc their first iteration looked a lot like this. I think they have since moved to a model of a few select locations where nodes are closer to each other.
This along with the increase in heat pumps, really makes me question air source heat pumps not directly hybridizing with water heaters (new water heaters have heat pumps that will cool your utility closet).
But using house heat pumps to cool air / heat water in the summer could at least decrease energy use during the day; and then in the winter, the water (as a heat source) can help assist older heat pumps in pumping heat from the cold (though modern heat pumps solve this problem)
in eastern europe they pump hot water directly to homes . it’s waste water from nearby power stations . you could do the same from the data center to homes
most datacenter's will recycle their water. Heat it up, send it somewhere else to cool down, send it back to heat up. Specific datacenters were called out on here last week for dumping heat waste water. The efficiency of moving hot water over long distances doesnt make much sense
my point is that it's commonly used e.g. in Finland & Russia, so someone has figured out how to get hot water from outside the city into the city and heat homes.
In Germany the startup that became Cloud&Heat Technologies started with the idea to sell a heating system for home powered by some computers. It seems that they have pivoted to sell 19" rack with water cooling that can be connected to a building heating system, more for companies than individual household.
It looks like they didn't got much interest in the market and pivoted more to industrial customers. The main purpose now is not a computer node as a heating, but instead recovering the waste heat from computer rack for the heating system.
I wonder if you could do this yourself with a heat exchange (like from McMaster-Carr).
I've read before that if you run the loops "backwards", such that the two loops of the heat exchanger are flowing in opposite directions, that you can increase the rate of heat transfer by maximizing the difference in temperature.
This is called a counter-flow heat exchanger or countercurrent exchange, it's very commonly used in industrial applications, for both gas or fluid flow.
This setup is very efficient at heat transfer, and can counter-intuitively produce a higher output temperature than the counterflow's input temperature.
It heats the water tank, just as people heat their water tanks now in the summer. No difference really, except some of that energy is free to the homeowner.
Natural Gas companies near me began offering fiber services before Google Fiber did; because they had the infrastructure to monitor the lines and distribution nodes;
Given that fiber setup, the gas company has a low latency network.
From the FAQ on the trial page, it seems like it will simply not run if there is nowhere to dump the excess heat:
How much of my hot water will it deliver?
A useful base load. If the heata unit ran all day it would deliver about 4.8kWh of hot water which is about the amount of hot water an average household uses. In practice the utilisation level will be lower and sometimes the water will be at temperature so the heata unit won't be able to run. We think it could deliver up to 80% of an average household's hot water, with good compute utilisation. This trial will help us learn more about how this will work in practice and enable us to give you a useful amount of free hot water at the same time.
My solution? Two giant rack mount servers, running Foldign@Home at full capacity. Way I figured is that it's still 100% efficient at generating heat, but I might as well help drug research in the process.