Slashdot Mirror
Servers Ahoy — Startup To Build Floating Data Centers
1sockchuck writes "Startup International Data Security says it is moving ahead with plans to build data centers on cargo ships docked in the San Francisco Bay. IDS first announced its plans in 2008, but they were postponed by the credit crunch. The company says it has now lined up funding and an anchor tenant for a proof-of-concept 'dataship' that will hold 500 racks of servers in its cargo holds. IDS isn't alone in contemplating ship-board server farms, as Google has applied for a patent for a 'water-based data center.'"
Wonder if data sinks will have to be prohibited?
--- Users are like bacteria -> Each one causing a thousand tiny crises until the host finally gives up and dies.
These guys better watch out for software pirates!!!
so will this be an opportunity to pirate data in a whole new way?
Does this mean that file sharing and sites like Wikileaks could just pull up anchor and go to the next country if being pressured by local law enforcement? Interesting the possibilities that this could have. Can think either bad or good, maybe even both.
Seriously? How is a server-farm in a ship innovative enough for a patent? Goodness.
"There is a way that seems right to a man, but its end is the way of death." Proverbs 16:25 (NKJV)
You sunk my Exchange server! (THANKS!)
Never underestimate the bandwidth of a cargo ship full of servers, hurling through the Pacific ocean...
Kid-proof tablet..
Those servers are going to have CPUs inside them, so why not go with the Chips Ahoy! joke?
As an aside, that link is the worst possible way to do an URL for a website, especially a commercial one. You shouldn't need a redirect to be able to print your URLs in commercials and newspapers. Learn to do URL rewrites, it's transparent to the end-users.
So what are the laws concerning data resident on vessels under a foreign flag ?
Seems like a nice way to get pirate-bay like content really close to the US backbone.
Your server will be dry docked while repairs are made to the hull. Expected maintenance window: Fall 2011 -> Spring 2012
But it sure seems like a tsunami would take it out.
how long do you think it will be before the thermal pollution watchdogs start cracking down?
will make data centers as we know them obsolete. i'm already carrying a terabyte around in my pocket.
They still need massive data and power lines coming from the grid, and because servers need to be connected to the internet without even the slightest interruption, a floating server rack cannot be mobile. In fact, special steps would have to be undertaken to make sure it stays in one place during storms and other maritime crises. Wouldn't it make more sense to just buy a piece of land near the sea and simply pump the ocean water around for cooling? Throw in a few photovoltaic cells and a wind turbine and you'd get a far cheaper, more reliable land-based data center.
I assume they are using water-cooled CPUs?
I've abandoned my search for truth; now I'm just looking for some useful delusions.
..can't trust the floating point calculations, which have assumed dramatic new importance
I have been a user for about 10 years. This ends Feb 2014. The site's been ruined. I'm off. Dice, FU
Why on earth would you do this? This seems like a disaster waiting to happen... Why not just put the servers in the sewer or dangle them from blimps? Put them in orbit? How about in Pakistani caves? Submarines? On the backs of sharks? Whalesharks?
My "water based fishing center" will use a complex system of nets, cranes, and a crew to fill these ingenious live wells with the bounty of the sea. Anyone who uses my technology will have to pay me a LOT of money to license it.
Ships aren't cheap, and marine environments are rather hostile (salt, water), and data centers can already be reasonably mobile by putting it in a shipping container and moving that shipping container somewhere... so what need is this filling?
is competition good, or is duplication of effort bad?
I bet you could get some super cheap cooling just by puttering up into the arctic ocean and opening a few windows. Solves one data center issue at the cost of others.
I wonder though if anyone has thought of building land based data centers in the far northern climates to take advantage of the -40 degree arctic cold fronts.
I used to joke with a friend who wanted to do a tour in one of the antarctic research stations that he could leave a overclocked computer sitting outside his bedroom window, and the only problem he would have is penguins trying to lay eggs in it.
HA! I just wasted some of your bandwidth with a frivolous sig!
So when are ya'll gonna start making Piratbay jokes?
of a cargo ship full of servers?
From a purely technical standpoint, I'm still not seeing the benefits of a ship-borne data center. For such a system to be useful, you still have to connect it to the shore somehow, and you still need a significant power source, both of which rather necessitate tying your data center up to a dock. As far as cooling with seawater, that's a nifty idea, but one that could just as easily be done by running some pipes from the shoreline.
So that suggests that the motivation for this isn't technical at all, but legal. As in "we need a way to get all our stuff into international waters relatively quickly". Which means these are not financial backers or "anchor tenants" I'd want to have anything to do with.
I am officially gone from
Han Solo. I'm captain of the Millennium Falcon. Chewie here tells me you're lookin' for 26 racks plus eight 40 amp drops?
Obi-Wan: Yes indeed, if it's a cool ship.
Han Solo: Cool ship? You've never heard of the Millennium Falcon?
Obi-Wan: Should I have?
Han Solo: It's the ship that made the Kessel Run at less than twelve degrees. I've out-chilled Liebert refrigerators. Not the local bulk chillers mind you, I'm talking about the big water-fed units now. She's cool enough for you old man. What's the cargo?
Obi-Wan: Only bladecenters. Plus two droids... and no questions asked.
Han Solo: What is it? Some kind of torrent tracker?
Obi-Wan: Let's just say we'd like to avoid any American entanglements.
Ship, noun: "A hole in the water into which one throws money."
And let your data center run into another ship....
http://online.wsj.com/article/SB10001424052748703428604575419212377490340.html?mod=googlenews_wsj
So I'm guessing they have to dock these things and keep them hooked up to the power grid. Running a datacenter off of a deisel engine sounds exspenive and impractical. Probably don't want to explain to your customers that their DC is down because it ran out of gas. So that kind of cancels out the mobility idea once the thing is up and running doesn't it? But maybe building these as ships first can proove the concept and then they can build them as sea platforms later?
enough lifeboats.
Obvious serious downsides (No pirate jokes)
1. Less bandwidth capacity - higher latency (forced to use sat. solution)
2. Very very non-green. Ships put out awful emissions. One running tons of electrical equipment is no better.
3. Curious to know if the sea's movements will affect disk latency - tests have been conducted that if you scream at a hard drive it can slightly affect their latency due to the sound waves.
4. International waters. Nuff said.
5. I'd be willing to say a ship crew - security, EXPENSIVE ship repairs and maintenance far outweigh the profitability.
Think of the ships on 'Deadliest Catch' during a storm.
Do you want your data stored on those platters? Pitching / Lisping 30 degrees or more? How is that handled?
Presumably one advantage would be that you could register your "ship" in Liberia, Panama or some other flag of convenience such that you pay minimal taxes, avoid local labor & environmental laws, etc.
Support Right To Repair Legislation.
lol, GET IT?
I've come to an idea of placing data centers in proximity to trans-ocean cable landings on ocean floor. Crazy, huh ? :) :)
It might be started at the border of neutral waters, around 5-10 km from shoreline and later expanded along major submarine cable systems.
Perhaps, initially it might be non-maintainable sealed containers with racks, powered by either
a) elements using natural forces like waves or streams
b) galvanic elements
c) hydrogen (there's definitely no lack of water)
and cooled by external medium (also water).
I'm planning to present this approach at UP 2010 Conference in November 2010, so stay tuned.
As cool as this sounds in a Gibsonan sense (or Stephenson if there is a guy named Raven with a nuke and glass harpoon running around) how are they getting the bandwidth? It seems like that cost would outweigh gains from conventional data centers.
"I'm not a quack, I'm a mad scientist! There's a difference." - Dr. Cockroach
This might be flamebait my personal opinion is that this idea is abject stupidity. I don't see any benefits as the ship still must be moored for ground power. I could see this being really good economically if you could anchor the ship in a place where you could harness waves for power. Perhaps, the only benefit is mobility - you could move the data center to where it is needed. However, with the urban blight resulting from the latest economic meltdown, why not grab up some of the abandoned buildings, rehabilitate them, and turn them into working datacenters thereby creating jobs for other industries and making America look better.
I think this could work on some sort of Oil Rig type Platform; however then you would have to fly everyone in and out via the Google Shuttle helicopter. Actually it is kind of Google's dream to have everyone live at work, thus their laundry, napping, etc....If everyone works on the rig then they can't have anylife outside of work, starting to feel real Matrixy/China.
"System, Method, and Apparatus for sea-based collection of oceanic animals consisting of a collection apparatus, a temporary ocean-like storage environment, transportation vessel, and a system and method of moving said animals into and out of the temporary storage environment."
that computers and water don't like each other?
> powered by [...]
> c) hydrogen (there's definitely no lack of water)
Because in neutral waters we don't have to obey the laws of thermodynamics.
Could you explain, please?
It takes more energy to split hydrogen and oxygen apart than can be recovered by putting them back together again.
Alexander Peter Kristopeit bought his basement from his mommy for one dollar.
Basically: Water is burned hydrogen. Mentioning it as a source of chemical energy is nonsense.
You need energy (eg: electricity) to turn water into hydrogen and oxygen (and heat).
Hydrogen and oxygen of course can be used to generate electricity (and heat and water).
In the end you'll have less electricity than you put in but way hotter water.
(I'm amazed that I'm explaining this on slashdot)
What's that? What are all those planes in the sky heading this way???
The tyrant will always find a pretext for his tyranny - Aesop
Surely it would be better to tether a submarine just outside territorial waters and have it link to the mainland via microwave relayed via a tethered blimp or via undersea fiber. The submarine could have a couple of access tubes leading up to the surface for air, access and a place to tether the blimps. That way, there would be almost no "wave action" to rock the datacenter.
Nullius in verba
Well this is good point, but this way you can negate anything :)
What I was thinking about is that water could be broken into hydrogen and oxygen by electrolysis—a photoelectrochemical cell (PEC) process which is also named artificial photosynthesis. Research aimed toward developing higher-efficiency multijunction cell technology is underway by the photovoltaic industry.
Here's process description from public source:
Electrolysis of water
Hydrogen can be made via high pressure electrolysis or low pressure electrolysis of water. In current market conditions, the 50 kWh of electricity consumed to manufacture one kilogram of compressed hydrogen is roughly as valuable as the hydrogen produced, assuming 8 cents/kWh. The price equivalence, despite the inefficiencies of electrical production and electrolysis, is due to the efficiency of direct conversion of fossil fuels to produce hydrogen, rather than burning fuel to produce electricity. However, this is of no help to a hydrogen economy, which must derive hydrogen from sources other than the fossil fuels it is intended to replace.
Here's a working storage http://en.wikipedia.org/wiki/Hydrogen_Challenger
High-pressure electrolysis
High pressure electrolysis is the electrolysis of water by decomposition of water (H2O) into oxygen (O2) and hydrogen gas (H2) by means of an electric current being passed through the water. The difference with a standard electrolyzer is the compressed hydrogen output around 120-200 Bar (1740-2900 psi). By pressurising the hydrogen in the electrolyser the need for an external hydrogen compressor is eliminated, the average energy consumption for internal compression is around 3%.
No need to be amazed :)
See my response above with working mobile production in open sea http://en.wikipedia.org/wiki/Hydrogen_Challenger
also note launch year, it is 1967, so process of generating electricity for the electrolysis of water is pretty mature :)
These only can be truly mobile with nuclear powered generators and dedicated satellite for each boat. Good luck with that!
I'm a maritime professional (captain of a ship) w/ some collateral involvement in IT issues. Here are a few off the cuff thoughts / questions that spring to mind as I read /. and eat my lunch:
1. Environmental control . . . a bit more difficult to do in a big metal box that is your modern ship, especially in a hot sunny place (e.g. San Diego). Ships can also be very damp inside. Proper environmental control for a data center on board ship will almost certainly be more difficult than for a land-based data center.
2. Watertight integrity of the hull . . . ship is floating in a harbor, bay, ocean that is exerting constant pressure to swap places with the air inside the hull (i.e. flooding). Ships have a number of through hull fittings / valves that need fairly frequent inspection / maintenance to ensure proper operation and guard against leaks.
3. Crew . . . While I'm sure the DBA and all the other information technology professionals are smart guys and gals, they are most likely not even remotely qualified and/or licensed to sail the ship. The more competent the crew you want to sail the ship, the bigger the bill. You don't need them aboard all the time, but you do need to get them in time to move the ship away from an impending disaster, political unrest or whatever motivates you to sail the ship from where it is presently moored. There will likely be other complications with the crew, depending on where you have it registered. Even if you don't maintain a crew on board to sail the ship, you still need a caretaker crew of some size to look after the ship on a daily basis.
4. Maintenance . . . Ships require lots of care, even if they are just sitting in the same mooring without moving for years at a time. Caring for ships is a spendy proposition as noted by one reader already (hole in the water into which one pours money is a fairly accurate statement). Keeping the ship continuously ready for sea is even more expensive. Someone has to exercise the main engine, generators, deck equipment, etc.
5. Data Connection . . . Obviously the ship will need lots of land line connectivity. not an insurmountable problem pierside, but out at sea . . . very, very big bills for relatively small bandwidth connections. Would need large / expensive satcom installation / subscription for even a rudimentary at-sea high bandwidth connection.
6. Pier Space . . . ain't cheap.
7. Building / converting the ship. . . . Having a ship built or converted for a special purpose such as this will result in an eye-popping bill (and that will be in addition to the purchase price for the ship itself if you are buying an existing hull).
All in all, save for the mobility gained by being able to physically sail away to another port with your server farm, I can't see the benefit from doing this.
I guess the term "Anchor Tenant" has multiple meanings in this case. I wonder why they're doing it in San Francisco, with it's incredibly restrictive business rules and ridiculously expensive and unreliable electricity?
gather all the seamen to the poopdeck! think that's bad? they even named one of their servers swallow
Other than 'land is expensive'
Boats are expensive too. Very expensive actually and require a lot of labor to maintain. Not to mention that water and electronics are not comfortable bedfellows. Plus you need an inexpensive source of electricity. Marine diesels are efficient but still quite costly compared to the grid. I can think of good reasons to have a floating data center but reduced costs doesn't strike me as probable.
A ship staying afloat on water.
On one case we have a company promoting deep underground storage - protection against the elements, etc. and on the other case we have one building storage on a ship floating on water.
Maybe there's a reason why this hasn't been done before - because it's moronic?
I really hope none of my money is being invested in this.
Sealand ....
International waters I can understand; but wouldn't it make sense to use a bay or dig yourself a bay off the coast? you get water flow from the ocean and don't have to worry about storms or much motion.
Democracy Now! - uncensored, anti-establishment news
Couldn't they find a better acronym than the one for Sudden Infant Death Syndrome?
than burning the hydrogen will give you. That's what I tried to explain.
So the hydrogen in the water will not help in any way to supply the server ships with electricity.
he/she mentioned the hydrogen in the surrounding water as one potential source of electricity.
It's not, obviously, because it always takes more energy to split water in hydrogen and oxygen than you can get in electricity from burning the hydrogen again. The rest is heat. That's what I wrote and what I was amazed at having to explain on Slashdot.
So the water around the site does not solve the problem of how to supply the site with electrical energy, at least not by using chemistry.
Of course there may be options to generate electricity on site using wind/wave/solar but you would not want to waste it on generating hydrogen and heat first but to power your serverfarm, directly. Maybe you would want to use it to generate a litte hydrogen for USV-purposes, now that I think of it, but still.
Not anything; only thermodynamically infeasible things.
Without adding power from some other source, you cannot split water and then use it for energy. It costs more energy to split the water than you can get back from the hydrogen when you burn it. Your combustion engine is limited by the Carnot cycle.
If you have power from some other source, you’d waste some of it if you used it to split water to get hydrogen. Although it is still conceivable that the net efficiency of using the available energy to split water and harvesting the hydrogen for fuel would be more efficient than your means of using the available energy directly.
Alexander Peter Kristopeit bought his basement from his mommy for one dollar.
Seriously Slashdot, for shame!
The RIAA wants Pirates I say we give 'em pirates! Just imagine the fast sloop Datahawk closing on the large and unwieldy Datastore, just watch out for the HCMS Defender. (I have no idea what I went with all "D's" there). Anyway Heave to and have at her! Full broadsides, get that connection going!
Dress Code for the Sys Admin includes a eye patch and a peg leg. As you were for personal hygiene.
If they dropped their anchors somewhere in polar circle instead of SF, they would save huge money on cooling. Propably savings would be enough to fund longer power and fiber lines.