It takes more work to avoid salt buildup, but you can evaporate saltwater as a place to dump heat, and we aren't gonna run out of saltwater any time soon. 'Course, only so many places have saltwater access.

EDIT: You evaporate enough water for cooling, you can increase rainfall somewhat in the local area, which boosts crop growth measurably. I remember reading an article about nuclear power plants that use evaporative cooling producing that effect.

kagis

Effect of Nuclear Power Plants on Local Crop Yields | Journal of Agricultural and Applied Economics | Cambridge Core

Effect of Nuclear Power Plants on Local Crop Yields - Volume 54 Issue 1

Cambridge Core (www.cambridge.org)

The growing prevalence of clean energy raises the question of possible associated externalities. This article studies the effects of nuclear power plant development (and, as a result, the increased amount of water in the atmosphere from evaporative cooling systems) on nearby crop yields and finds that an average nuclear power plant increases local soybean yields by 2 and corn yields by 1 percent.

The problem with using salt water isn't salt buildup, it's that it's corrosive and will drastically shorten the lifespan of any equipment exposed to it.

The problem with using salt water isn't salt buildup, it's that it's corrosive and will drastically shorten the lifespan of any equipment exposed to it.

What do datacenters need large volumes of running water for? Can they not do a continuous loop? It's for cooling computers, right? That can't be done with a closed loop of water?

A closed loop just moves away the heat, you still need to cool the water, else it keeps rising in temperature until it doesn't cool enough. On desktops this can be achieved with fans, which isn't surprising as waterless setups of fans can already cool down most desktop CPUs.
On a data center (and power plants), this is not feasible as they generate too much heat. They would need massive fans and would raise the air temperature like crazy.

My next thought would be to find a use for the heat; colocate the data center with some other facility that needs massive quantities of heat. I remember something about a spa that heated its pools with computers (I think mining bitcoin, but still). I've also been curious if heat pumps could get hot enough to bake bread at industrial scales. Pump heat out of a data center and into a bigass bread oven?

How about reducing our dependence on data centres by using software that is more peer to peer and local first etc?

Of course some data centres have legitimate use cases, such as big data analysis on weather and climate data etc, but building huge data centres for social media and running everything in the cloud is silly from an environmental perspective

How about reducing our dependence on data centres by using software that is more peer to peer and local first etc?

Of course some data centres have legitimate use cases, such as big data analysis on weather and climate data etc, but building huge data centres for social media and running everything in the cloud is silly from an environmental perspective

Distributed computing would eliminate the water usage, since the heat output wouldn't be so highly concentrated, but it would probably somewhat increase power consumption.

In an ideal world I think data center waste heat would be captured for use in a district thermal grid / seasonal thermal energy store like the one in Vantaa.

Of course that isn't to say that we shouldn't be thinking about whether we're using software efficiently and for good reasons. Plenty of computations that take place in datacenters serve to make a company money but don't actually make anyone's lives better.

How about water usage rates that penalize bulk consumers instead of giving them cheaper rates?

Also, good luck sourcing salt water in Illinois.

Distributed computing would eliminate the water usage, since the heat output wouldn't be so highly concentrated, but it would probably somewhat increase power consumption.

In an ideal world I think data center waste heat would be captured for use in a district thermal grid / seasonal thermal energy store like the one in Vantaa.

Of course that isn't to say that we shouldn't be thinking about whether we're using software efficiently and for good reasons. Plenty of computations that take place in datacenters serve to make a company money but don't actually make anyone's lives better.

If I'm having a video meeting p2p instead of microsoft teams running in the cloud, that would reduce power consumption, not increase it.

It takes more work to avoid salt buildup, but you can evaporate saltwater as a place to dump heat, and we aren't gonna run out of saltwater any time soon. 'Course, only so many places have saltwater access.

EDIT: You evaporate enough water for cooling, you can increase rainfall somewhat in the local area, which boosts crop growth measurably. I remember reading an article about nuclear power plants that use evaporative cooling producing that effect.

kagis

Effect of Nuclear Power Plants on Local Crop Yields | Journal of Agricultural and Applied Economics | Cambridge Core

Effect of Nuclear Power Plants on Local Crop Yields - Volume 54 Issue 1

Cambridge Core (www.cambridge.org)

The growing prevalence of clean energy raises the question of possible associated externalities. This article studies the effects of nuclear power plant development (and, as a result, the increased amount of water in the atmosphere from evaporative cooling systems) on nearby crop yields and finds that an average nuclear power plant increases local soybean yields by 2 and corn yields by 1 percent.

Hmm. If a city had some system of central heating, would a data center's waste heat be used during winter months? I've heard of projects to use abandoned mines flooded with water as a sort of thermal battery, could that also be a solution?

Is a connection between 3+ people still p2p? Or is there another term for it?

I don't know how this would work over the internet though.

On a LAN you could use multicast, but I don't think ISPs support multicast, it seems like it would be an easy way to DoS. But I honestly don't know.

So, if you can't multicast, the way to have serverless multi-user video calls would be to have a separate video feed for each receiver, which I can see using more resources than through a server that would replicate the stream to all the receivers. Of course this is dependant on distance, even without multicast it consumes more resources if everyone is in the same LAN.