Slashdot Mirror
How Do I Put Unused Servers To Work?
olyar writes "I worked for an internet start-up last year and during the 'we have plenty of money' phase, a lot of server hardware was purchased. Eight months later, there is very little money, but we're still plugging along — using only a fraction of the hardware. We just cleared out a co-lo and I now have a stack of 17, 1U servers in my garage. Each of those has 2 servers, each of which is a 2-processor, dual-core box with 8 GB of RAM. Add that up and I have 136 processors and 272 GB of RAM with nothing to do. The IT guy in me thinks that's a waste of FLOPS. The wanna-be businessman in me thinks its probably a waste of money as well. So I've been brainstorming ways to put all of that power to good use. Any ideas?"
Yeah was thinking the same, smartest thing to do is sell it, way too much hassle to try and compete with existing services.
Start competing with your employer. If they can afford to do whatever it is they do, and still just give away thousands of dollars in gear, there's obviously room for improvement.
It takes a decent amount of electricity to run that much hardware. That may not be the kind of "donation" the OP had in mind - donating to the local power utility.
Donate them to Pirate Bay. If nothing else, it will help them with the streaming video for their trial. :)
Or, you could run Crysis in software rendering mode.
Or rent it out to spammers, crackers, etc.
Seriously, though....you could probably rent out time on it to researchers for less than most supercomputer time costs. Especially since the hardware costs you nothing. All you have to pay for is power. Figure out how much it uses running full tilt, double or triple that cost, install Linux on the thing, and rent out CPU time.
Maybe you could even be part of the next big breakthrough in security research.
"City hall" in German is "Rathaus" Kinda explains a few things......
Somebody gets cheap hardware, your company gets more cash, the servers get used for something worthwhile...everyone wins.
Even if he tried to do something profitable using them, between paying for power and bandwidth to operate them, it would have to be a real business model to even expect to break even in the modern economy of cheap professional server hosts. If there is a local university by you, I'd advertise trying to donate it to a local college or University with engineering/computer science programs. Often students just need academic clusters for the experience of parallel programming problems, and of course it could even help in minor actually useful research. And I'm sure they could help you work out a way to get some sort of tax recognition for the donation.
The IT guy in me thinks that's a waste of FLOPS
It's not a waste of FLOPS. There are plenty of spare MIPS and FLOPS in the world - witness the amount that get donated to folding@home, seti@home, various cipher cracking contests, etc. While you too could donate to those causes, I'd suggest against it - it's one thing to donate niced cycles of a machine that otherwise has to be running, but it's a tremendous waste of power to spin up that many boxes just to hand out cycles.
Recognize those servers for what they are - a waste of *money*. You sunk too much cash into a resource (and that's fine, no business has perfect foresight, and you had to anticipate potential needs). Now liquidate them and get your money out so you can spend it on something better than depreciation. If it turns out you need them in a year, I assure you you can buy servers for less $/FLOP from the liquidators at that time.
I'd like to take this opportunity to thank whoever coined the word 'Beowulf' as a very convenient shorthand form of 'I don't know anything about cluster computing, please disregard my opinion.' I can't begin to imagine how much time this has saved people over the years.
I am TheRaven on Soylent News
It's the perfect time to adopt a mad scientist. Seriously, how cool would this be to a Neural Net researcher? I honestly think you should put an advert out saying "looking for researcher to utilize private cluster with 272gb ram and 136 procs".
http://soylentnews.org/~tibman
Buying computers is a cost sink. You buy computers and amortize the cost of them over a few years. You ONLY buy computers because you need to do computing work.
If you don't need the computing power, sell off 90% of them at a great price (maybe 20% below market value), RIGHT NOW. Holding onto depreciating assets with no return on them is no better than tossing money into a furnace.
Keep a couple of 'em around for growth, spares, and new projects. Sell the rest, and when you need the computing power, buy something 'x' times faster for the same amount of money.
"People who do stupid things with hazardous materials often die." -- Jim Davidson on alt.folklore.urban
If you dont have a pressing need, why bother powering them up? Theres a reason $job pulled them from the colo. Im sure they are hungry heat-factories. Keep that green in your wallet when its time to pay the monthly power utility bill and just dont power them up.
And if youre not going to spin them up, punt them to someone who needs/wants them while they still have some value before Moore renders them useless ...
Regardless of what you do as far as utilizing the servers, call your insurance agent straightaway and make sure that equipment is insured! Business property is very unlikely to be covered by your homeowner's policy so theft/fire/whatever could leave you financially exposed (or even liable, should the investors choose to come after you for reparations).
You mean ::1
Not to burst your bubble on this, but the chances that you'll even come close to breaking even are highly unlikely.
First off, even if you're only running 500watts in servers, which is not more than 2 decent ones under moderate load, you are paying about $450/yr to keep them on 24/7. Assuming you're lucky enough to get power as cheap as $0.10/kWhr.
Unless you have some source of customers, i.e. you are a web-developer, or have lot of friends who want a sub-par web host, you'll need to advertise. The hosting market is incredibly saturated and Adwords on hosting keywords is very expensive. Expect to spend a few thousand per month for a few months to recruit your initial 100 users.
Once you've got 100, you can probably turn it down to a few hundred per month to keep to user count flat to make up for those you lose in turn over.
This also assumes that with that advertising you are actually able to sucker some users to pay $10/mo for static hosting on your cable modem, while they can get it for free from a number of providers like Google, or pay what amounts to a few bucks a month for hosting with an SLA, a real support staff and some level of redundancy and backup. There are a lot of suckers in the world, so we'll assume this is possible.
Now, after you factor in the time you spend answering inane support e-mails from your customers, you'll see it's probably more profitable to get a second job at McDonald's. If only flipping burgers was as much fun as playing with servers. ;)