Slashdot Mirror
Startup Builds Prototype For Floating Data Center
1sockchuck writes: California startup Nautilus Data Technologies has developed a floating data center that it says can dramatically slash the cost of cooling servers. The company's data barge is being tested near San Francisco, and represents the latest chapter in a long-running effort to develop a water-based data center. Google kicked things off with a 2008 patent for a sea-going data center that would be powered and cooled by waves, conjuring visions of offshore data havens. Google never built it, but IDS soon launched its own effort to convert old Navy vessels into "data ships" before going bankrupt. Nautilus is using barges moored at piers, which allows it to use bay water in its cooling system,eliminating the need for CRAC units and chillers. The company says its offering may benefit from the growing focus on data centers' water use amid California's drought.
>> growing focus on data centers' water use amid California's drought
Um...what? Don't they just chill the water, let the data center warm it and then reuse it?
Why not check to see what California agriculture's doing with it's majority share of the water first?
Is anyone considering the local effects of warming the water in the harbors these centers will be docked in? It seems to me, given the current toxic algal bloom off the west coast of the US at the moment, we might be just a bit concerned, right?
"Life is not magic." Dr. Ron Weiss - "If we don't play God, who will?" Dr. James Watson
Great for rusting
Someone you trust is one of us.
I can see it now, actual pirates stealing full boatloads of servers.
excitingthingstodo.blogspot.com
I'm skeptical that even at SF's inflated real estate prices that floating servers on a boat is cheaper than a ground-based datacenter. Marine structures are expensive to build and maintain and they have to pass regular USCG inspections. For cooling they could rent a warehouse near the bay and pump the water in.
Is it really cheaper to build a barge than it is to circulate sea water to a land based facility?
Good: more efficient cooling, which is better for the environment
Bad: ... but not if power is supplied by inefficient on-board generation
Ugly: Piracy
Serious problem... Redundant high speed Data connections...
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
It will be cheaper until the run smack into the environment regulations that limit how much you are allowed to heat a natural body of water. A data center won't be as bad as a power station using direct cycle cooling, but put enough of these "barge data centers" together in a single location, I presume they will congregate in areas of cheap power and high local bandwidth availability right? And you will hit the limits. Then you have to use much less efficient air to liquid or similar cooling towers anyway, along with the inherently higher costs of floating structures on high value waterfront property.
There are a few issues with this idea:
1: Is the data center going to be on a ship? If so, will it be in US waters? If not it is free game for any naval force whatsoever who feels like boarding the data center. There are a lot of issues about maritime property rights that make property disputes on land look tame in comparison, especially in international waters.
2: Reliable power? Ships are not static objects.
3: Stability. The closest thing to a stable platform is the prison barge attached to Riker's Island, and that thing requires a naval staff, bilge pumps, ballast tanks, and many other items. If they don't do their job, the barge can sink. A data center barge only will be worse.
4: Marine environments are by far the worst environments to even think of putting a data center in. There is a reason why anything related to boating costs 10-20 times as much as normal landlubber stuff. For example, just attaching a lug to a battery wire requires a $4000 hydraulic crimp tool as well as proper shrink wrap insulation. Same job for a car? $10 pair of pliers can do it.
5. Data connections will be difficult, high speed ones even worse.
6. Provisioning and resupply of fuel, maintenance spares, personnel and food will be very expensive.
7. Emergencies will be hard to deal with, fires, accidents, taking on water etc..
8. Storms, wind, waves, etc will all cause you issues.
But, as you point out, on board power will be the largest cost issue they face and will make this very difficult to accomplish, even if they can get it from natural sources and not have to generate it from fossil fuels.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
They could use the "Free" heat to boil the water to make distilled water and sell the distilled water!
Obama's legacy: (N)othing (S)ecure (A)nywhere and (T)error (S)imulation (A)dministration
These are barges in a bay not ships as sea. So those are non-issues.
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
Bad: ... but not if power is supplied by inefficient on-board generation
The ships are moored to a pier, and using shore based power. Burning bunker fuel to generate electricity is not only inefficient, but at least in SF is illegal. Bunker fuel produces filthy high sulfur smoke, and the boilers are required to be shut down as soon as a ship is moored.
It's even better if you can find some water that already has to be heated. The college I went to looked at using pool water as their heat destination. Their calculations showed it could be 163% efficient. I don't think the backup data center it was designed for was ever built, but it still seams like a decent idea. http://www.calvin.edu/~mkh2/thermal-fluid_systems_desig/2010-data-center-seminar.pdf
I thought of this, too. Is there a way to recover some of the waste heat and turn it back into electricity?
I'll bet the discharge water won't meet environmental standards, even if it's the identical stuff taken from the ocean. There was one guy who had a business farming fish and he was using ocean water. His discharge was cleaner than the intake water, but it didn't matter; they wanted him to clean it even more. He ended up shutting down and moving the business to Hawaii rather than deal with the intransigence of the bureaucrats.
Nothing says rust like a steel barge that floats in salt water and breathes salt air.
It is perhaps worth adding that here in the Northeast there is a powerful movement towards reclaiming the industrial waterfront for parks and green space.
For the curious, 95 examples of used barges for sale:
The add copy should be read like you were shopping for a second-hand boat in a "Monkey Island" game. Used Deck Barges
California could use the waste heat from the servers for desalination.
So what does Guam have going for it?
It is already a fibre hub. http://www.submarinecablemap.c...
It is less than 150 km north of the bottom of a very deep ocean trench. https://en.wikipedia.org/wiki/...
It is politically stable because it is a US territory. https://en.wikipedia.org/wiki/...
Not high enough. Far better to use waste heat from power plants to drive desalination
I prefer the "u" in honour as it seems to be missing these days.
Far better to use waste heat from power plants to drive desalination
Even better would be to forget about desalination, and just stop paying subsidies to people growing rice in the desert.
Cost of moorage, cost of running pumps to circulate sea water through your cooling system, cost of maintaining a boat, cost of maintaining the high speed data network to a boat. The only way this is cheaper is if you can't find cheap land near enough to the ocean to allow a pair of pipes to be run for cooling.
Yacht people sometimes describe them as holes in the water to pour money into.
Offtopic, and a bit more serious, but bouys on shafts using wave power make vastly more sense.
http://www.carnegiewave.com/projects/perth-project.html
The force of the waves pumps water and that can drive desalination as well as run generators.
Both may be necessary. At some point those making the decisions completely forgot that they should pay attention to reality and that there would by dry years every now and again just like through all of written history.
In fact, for AGW, we are supposed to see harder and longer droughts . the best thing is for America to prepare. With the waste heat from so many thermal power plants within 20 miles of the ocean, we are foolish to not use these.
I prefer the "u" in honour as it seems to be missing these days.
With the waste heat from so many thermal power plants within 20 miles of the ocean, we are foolish to not use these.
Modern desalination plants use reverse osmosis, driven by electric pumps, which have no use for "waste heat". Low grade heat could be used for the first stage in distillation, but distillation would still be prohibitively expensive.
LOL.
RO is used, but it is expensive. The reason why it is preferred is that it can be located anywhere and power lines ran to it. Now, if you have a power plant close to the ocean, esp if it is being used for cooling purposes, then desalination is next to NOTHING to do. Basically, it uses waste energy to create clean water. As such, it is NOT expensive, just limited in where it can operate.
I prefer the "u" in honour as it seems to be missing these days.
BTW, here is some information for you.
Basically, RO is expensive, but pushed by the wrong ppl. Thermal desalination using waste heat saves money.
I prefer the "u" in honour as it seems to be missing these days.
Moor it in San Francisco Bay, not dockside. Running power and network submarine cables a thousand yards off shore -- even redundant cables -- would be relatively cheap.... peanuts even, compared to the cost of Bay Area real estate.
This comment is my opinion and does not represent an official position of Donald Trump or others I do not work for
Can't we recycle the heat generated by data center to electricity?
Casteism
The problem is that you need shitloads of waste heat well over the boiling point in atmosphere - which is typically not wasted in such quantities. Everything that doesn't have a long and involved heat exchange system, using the boiler outlet water to do a LOT of preheating of the inlet water is either getting very old or is tiny - plus having to be coastal limits things a bit, so while there may be a few sites where it is possible they are going to be rare. There's even some places (eg. Liddell power station) where not only is the waste heat recovered as mush as possible but a bit of solar heating is added on top before it goes into the boiler to cut down on fuel use.
There's co-generation with stuff like burning methane from garbage, sewage etc with no boiler to heat, but projects like that are using their waste heat for stuff like warm water for aquaculture because they don't have the sort of waste heat to evaporate a lot of water - even at atmospheric pressure.
Gas turbines often have a bit of heat that's not recovered but you would need a lot of them at one site for this sort of thing, so there's an option for a new site designed for those two purposes.
So while it is far cheaper to use waste heat of you can get it there's not dependable amounts of it about for a total solution, and it's pretty expensive if you are burning stuff for the thermal desalination instead of getting the heat cheap as a by-product.
Thus if cheap heat is not available the opportunities shift to cheap mechanical energy to pump some pressure behind those membranes:
http://www.carnegiewave.com/projects/perth-project.html
these would be non-is
"5. Data connections will be difficult, high speed ones even worse.
6. Provisioning and resupply of fuel, maintenance spares, personnel and food will be very expensive.
7. Emergencies will be hard to deal with, fires, accidents, taking on water etc..
8. Storms, wind, waves, etc will all cause you issues.sues"
Okay 7 is still an issue but storms in a SF Bay, data connections, and provisioning would not be a big deal.
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.