Step 2: Service is subpoenaed. If the system is designed "corectly" there simply is not centralized directory. The information about who belongs to the system is not stored anyplace. Well actually it is distributed all over the system
Step 3: Certainly you could arrest people but proving they had any connection with the first person would be hard. They could rightly claim that their ISP was also in on it because the data went over their wires. It would be easy to show the subscribbers had no way to know what was on their machine
I'll donate 2TB of space each from multiple computers at different locations and between all of them i'm bound to have two critical pieces of your files, then all i have to do is shut them all down! Muah haha haaaa!
Your argument is that if the system were poorly designed then it might fail. Well "duh"
What if the files was distributed such that each block is stored in three places over a diverse set if IP addresses and then on top of that we compute a checksum block by taking the exclusive or (XOR) of each of group of blocks and writing out three copies of those. A system like this is very robust, more expensive but could survive even if MOST of the computers when off line.
This is how RAID5 works. On a raid5 system I can pull the data cable off any one of the drive in the array and still have no data lost. You shutting down a few computers would be like a drive failue in a large raid system. We can design the system to survive any number of failures if we have enough drives.
I can imagine where there might be a control where users get to select the level of security they are willing to pay for. The above system could still work if as many as 9 out of 12 computers went off line. People could "buy" that level of security in exchange for donating more disk space.
You don't need a "master server". There are lots of ways to get around this. One is to simply ask where your data is. Ask 10 other machines "Have you seen data item 45635? they in turn ask 10 other machines in about ten cycles every machine in the system has been quarried. This could work but would be a massive waste. But what if each machine cached it's queries? Then most could be answered without sending a message.
This is how DNS works. I ask my local DNS server "What is the IP address accociated with slashdot.com and it tells me. A distributed file system could answer questions like "What are the 4096 bytes on sector 4562354?"
Yes some machines might be down but we would store each sector three or more times.
The real problem is bandwidth. The cable the connects my hard drives is Firewire 800 and runs at 0.8 gigabits per second. My Internet connection is one ten thousand times slower.
The other argument that some one might but "bad stuff" on my disk and I'd be blamed. US courts have already held that the US Post Office and the Telephone company can not be blamed if people use their service to plan a criminal act. It would be easy to show that the data are encrypted and you can't have the key and that you don't know who else is using the system. But more then that you simply would NEVER have the complete file. You would not even have part of an image but only data that could not be used, even if you could decrypt it. You likey only have 2 or 3 random blocks of a 1000 block file.
they deliberately changed the driver model at the last moment so XP drivers wouldn't work?
Linux did the same thing. When the kernal version changed from 2.x to 2.y many drivers broke. But so what. they were fixed. Microsoft COULD have gotten it right. Other got it right even when they had to change the driver/kernal interface and even while suporting more hardware than vista.
"background task consuming enormous amounts of RAM and network bandwidth and otherwise misbehaves, it's going to make the experience shitty"
No, it does not have to be that way. Microsoft could have designed a kernel levl scheduler that always makes sure that the foreground task, the one the user is interacting with is responsive. These kinds of issues were worked out in the 1960's and are covered in university level computer science clases such as "operating system design 101".
Mac OS X is very good at this. I can write a resource hog app and un it in the background but at the same time iTunes will play music without skipping. Even on an old G4 machine.
You also can't argue that MS has to suport all kinds of different hardware. Linux and BSD Unix support a much wider range of hardware.
"Panels on the moon would require vastly more infrastructure to support them"
I disagree. They could be even lighter on the moon. For eaxmple make them paper thin and then unroll them and leave them face up on the ground. There is no wind to blow dust onto them like on Mars. You need nearly zero support structure
But the down side it they could not rotate to track the sun so you need a much larger array nd a much larger battery system because of the larger period of darkness.
My idea would be to use both. Let the nuke plant run the base loads and then do other things like charging the rover's battery or driving robots trackers, running pumps and so on when the sun is up. That way you don't need the huge battery. The solar array would only be usfull about one wek a month but could be put to good use
What happens when it's dark? The moon rotates and it's dark for 14 days then light for 14 days. A solar power system would need batteries were large enough to last for 14 days. We are talking literaly tons of batteries that would have to be soft landed on tha moon and then assembled.
The solar array would have to be sized to about 4 times the desired output because the sun angle would moslty be non-optimal and it would have to use at last half the array's output to recharge the battery
Yes it could be done but it is worth doing a trade stdu to compare cost of each.
Re:Most expensive science experiment ever?
on
LHC Flips On Tomorrow
·
· Score: 0, Troll
No, the ISS can't be the most expensive science project because it has little to do with science. ISS is a project alright but not a science project. Kind of an anti-science project realy as it's main purpose seems to be to suck up money from science related stuff.
"Anyway, Tennis is a relatively trivial example but things that happen in the physical world where physical forces are in play do not tolerate internet like latency very well. You cannot send xon/xoff like flow control signals to reality."
That is correct, so the way it would work is to isue higher level commands. Much like a coach would to a player. The coach gives only higher level statigy like "Stay more left of center and move up a bit." As robots become better they will need les and less real-time control.
"...a dot in the south pacific? if one were given to random chance, that's a lot of wasted souls in outrigger canoes in watery graves...."
People who lived on those islands tended to multiply and then run out of food. So they kill each other and took over the the limited land. The prehistory of Hawaii is one of endless wars and food and land rationing.
What happened is that people had to escape certain death by taking to the ocean in slim hope of finding another place. Over the centuries one in a thousand did find that other place. Most did not take to the oceans and were simply killed and their land taken.
Yes. I think that's the only answer that fits. Whoever first walked to North America must have come from North Asia. So it appears that maybe the population of N. Asia had a mix of peoples in it
Blu-Ray will be the LAST physical media. Samsung knows this and is not investing in R&D for the next media format. They've given that up and are now working on the new big dis[play technology.
"..in standard formats (e.g. CompactFlash, SD), you can just copy them bit for bit...."
No. The card card contain a active cryptographic device. But yes you could copy the data bit for bit but all you'd have is the encryted data. The point of using flash is that every copy is different because every SD card uses a different key and the key can't be read out of the card.
Don't say it's only a matter of time until it is cracked. Only poorly designed systems have been cracked the good ones have stood up for decades.
"...The chunk that actually broke off was 10 times the size predicted. Not sure why the big difference,..."
Some things are simply hard to predict. For example, we can predict that if a brick is thrown at a glass window the window will shatter but no one can predict the exact location of all the cracks in the glass in advance.
The falling leave problem is the same. We know it will fall down but can't predict the exact location.
All of this kind of stuff happens because very small, to small to measure initial condidtions cause huge changes in the result.
Recycling (except for aluminum cans and papers) uses more energy and costs more than creating new material
No. Recycling steel is very cost effective. Most low grade steel, like rebar is made from recycled material because it is cheaper.
Plastic bottles are used in big numbers buy both industry in Koera and the carpet manufactures in the American South.
Green waste is finding good use too. The price of mulch has falled which means more is being used which means there is less need for water, insectisides and so on.
ANd then the biggest use of re-cycling is that land fills are not being filled. In most places it costs about $50 a truckload to dump trash in a landfill. Not including the cost of the fuel to move the trash. So even if re-cycling were not 100% break even cost effective the over all saving is great when you factor in the saving in the cost to dump in a land fill. In other works even if you only get $1 per truckload of plastic soda bottles that is better than paying $100 per truckload to put them in a land fill.
Since there's evidence of multiple cycles of warming and cooling on the planet, another reason might be that cycling warming and cooling is a normal pattern for our planet.
Yes this is also happening. What we see the the supper imposed effects of both. There are several effects all gouing on at the same time. To sort it out you need lots of data and good statistical technique.
My opinion of this is that because coal and oil are the cheapest form of energy people will continue to dig it up and burn it until such time as coal and oil become so scarce and expensive that no sane person would burn them. You can't fight greed. But the good news is that as China and others come up to Western standards it's only 50 years until oil will become to valuable to burn. I think the problem is real but will self-correct itself inside of 75 years.
You don't have the shred the entire drive. I've seen security guys disassemble the drive and place each platter on a belt sander. Simply removing the coating is enough. Welding torches do a good job too. the coating come off with enough heat too.
It all depends what's on the drive in most cases the over write the data 10 times with random data is "good enough" but with other types of data physical destruction is the only option. There are many, may agencies each with their own rules about this.
Your ball of foil was NOT a faraday cage. Or rather it was a very poorly constructed one. You need a fully conductive enclosure with very tight seems. In professionally constructed enclosures you will notice things like copper gaskets and closely spaced bolts, doors where a copper knife edge is forced by a cam into a narrow slot. Even after the enclosure is built it must be tested and small air-gaps found and repaired.
In your case, a foil to foil joint is likely not conductive. Oxide coating on the Al foil acts as an insulator and you get a few ohms of resistance, at least.
Try another test using copper or brass pipe. Put the phone inside a length of pipe and screw end caps over each end of the pipe. Tighten until the bare metal on metal threads from a gas tight seal.
Yes this can be done. It is done. But in engineering we typically work from a set of requirements. This means we get a spec that says the part must fit in a given space or use only so much power or whtever. What you have is a solution looking for a requirement. In other words "I want to epoxy pot my M/B - Why would I do this"
If you must use this type of cooling why not simply us a nonconsecutive liquid? If you potted the M/B you'd have to pot all the conctors too. You never be able to replace a cable or RAn or move a jumper
The other thing that engineers get paid to do is come up with cost effective designs. In terms of cost this is just the wrong way to go.
The best way if you wont a compact and silent machine is to use the case itself as a heat sink. Conduct the heat to an aluminum case and let it radiate. Use a termally controlled fan as a backup for hot days and high loads. Oh wait isn't that what Apple does?
I don't buy the argument that Windows is targeted because it has 90% market share.
Do termites "target" wood frame building and avoid concrete structures because 90% of buildings are wood? Termite don't bother to each bricks because there are not so many bricks?
Same here. virues "eat" Windows and not any of the other OSes because Windows _can_ be eaten and the others can't. Windows is a soft target it was designed with the idea that it would run on an isolated desktop with no networking. Most all the other OSes are UNIX based. Unix was designed, back in the late 60's when computers were shared by many users. Unix was designed from the start to keep users isolated and unable to efect each other.
Yes Windows could improve but if it were to loose it's backwards compatabilty Microsoft would loose it's monopoly
What he's saying is that is appears to be written by some low-skill windows only programmers. That is the problem with open source. The whole world will see your work and judge you by it. Appearently this was written by some people who didn't know there was a world other then Windows on a PC.
I bet you can't read your doctor's prescription slips either. Every indutry has it's own set of terminology.
Why Chrome? If you are only viewing today's web pages then today's browsers are "good enough". But what Google hopes is that in the future you will be runing Google's on-line office suite, Google Maps and Google Mail. With Chrome all of these Google apps will run in their own tab, in their own thread and on one of the four or eight cores in your CPU.
Google knows full well that it takes many years to develope a mature web browser. This is their first step.
Today people don't run three or six different javascript aps all open at once. But Google, with Chrome hopes to change that.
This is related to the reason some people say "I don't need an 8 core machine". Of course that is true because you are running software that does not need 8 cores. But in 10 years your software will do different tasks. Maybe you will control the computer with hanf movments in front of a stereo 3D web cam while another camera watches where your eyes are looking, so as to update that part of the screen first. Same with browsers. Google is thinking ahead five to ten years.
"The show-stopper is(as of now) no NoScript/AdBlock! I've become spoiled with ad-free pages"
There are other ways to block ads. I run an HTML proxy server that also funtions to remove ads and other "junk". The proxy allows me to run any browser I want. If you have several computers the proxy works even better because it can block ads for an entire LAN and also cache common content.
orbital trajectories pretty well known, yes but hard to predict out to the future. So while you might know exactly where something is right now what you don't know is where exactly is will be next week. The unknowns are things like how much drag might slow it down, the atmosphere is not 100% constant and we don't know the shape of the object. Earth's gravity varies be location. We don't know about Solar activity and then there is the gravitational effects of the sun, moon and planets (Jupiter). With so many very small effects that can combine and interact there is a certain amount of error in any prediction
You have a room full of computers and you can afford $600?? Something is wrong.
Is the room full of cheap desktop PCs? if that is the case then your problem is that you yu are using to much power. Replace and consolidate the servers
Back to the "can't afford $600 problem: have you figured out what it will cost per month to run your $600 AC unit? I'm thinking that you can burn up $600 of power in 90 days. If you were to replace those servers you'd save a bundle. Look into something like one of Sun's "cool threads" servers. Typically one of these can replace a rack of PCs It's a very low power 16 core machine.
In the mean time it ooks like you've found your low cost solution, only that you are not only paying to much for power now but the $600 AC unit will double your power bill. Best to cut power usage with low power servers. Every wat that the server burns takes more than one wat to cool with an AC unit.
Slashdot Mirror
Two problems with the
Step 2: Service is subpoenaed. If the system is designed "corectly" there simply is not centralized directory. The information about who belongs to the system is not stored anyplace. Well actually it is distributed all over the system
Step 3: Certainly you could arrest people but proving they had any connection with the first person would be hard. They could rightly claim that their ISP was also in on it because the data went over their wires. It would be easy to show the subscribbers had no way to know what was on their machine
I'll donate 2TB of space each from multiple computers at different locations and between all of them i'm bound to have two critical pieces of your files, then all i have to do is shut them all down! Muah haha haaaa!
Your argument is that if the system were poorly designed then it might fail. Well "duh"
What if the files was distributed such that each block is stored in three places over a diverse set if IP addresses and then on top of that we compute a checksum block by taking the exclusive or (XOR) of each of group of blocks and writing out three copies of those. A system like this is very robust, more expensive but could survive even if MOST of the computers when off line.
This is how RAID5 works. On a raid5 system I can pull the data cable off any one of the drive in the array and still have no data lost. You shutting down a few computers would be like a drive failue in a large raid system. We can design the system to survive any number of failures if we have enough drives.
I can imagine where there might be a control where users get to select the level of security they are willing to pay for. The above system could still work if as many as 9 out of 12 computers went off line. People could "buy" that level of security in exchange for donating more disk space.
It could work well but it depends on the design.
You don't need a "master server". There are lots of ways to get around this. One is to simply ask where your data is. Ask 10 other machines "Have you seen data item 45635? they in turn ask 10 other machines in about ten cycles every machine in the system has been quarried. This could work but would be a massive waste. But what if each machine cached it's queries? Then most could be answered without sending a message.
This is how DNS works. I ask my local DNS server "What is the IP address accociated with slashdot.com and it tells me. A distributed file system could answer questions like "What are the 4096 bytes on sector 4562354?"
Yes some machines might be down but we would store each sector three or more times.
The real problem is bandwidth. The cable the connects my hard drives is Firewire 800 and runs at 0.8 gigabits per second. My Internet connection is one ten thousand times slower.
The other argument that some one might but "bad stuff" on my disk and I'd be blamed. US courts have already held that the US Post Office and the Telephone company can not be blamed if people use their service to plan a criminal act. It would be easy to show that the data are encrypted and you can't have the key and that you don't know who else is using the system. But more then that you simply would NEVER have the complete file. You would not even have part of an image but only data that could not be used, even if you could decrypt it. You likey only have 2 or 3 random blocks of a 1000 block file.
they deliberately changed the driver model at the last moment so XP drivers wouldn't work?
Linux did the same thing. When the kernal version changed from 2.x to 2.y many drivers broke. But so what. they were fixed. Microsoft COULD have gotten it right. Other got it right even when they had to change the driver/kernal interface and even while suporting more hardware than vista.
MS simply screwed up.
"background task consuming enormous amounts of RAM and network bandwidth and otherwise misbehaves, it's going to make the experience shitty"
No, it does not have to be that way. Microsoft could have designed a kernel levl scheduler that always makes sure that the foreground task, the one the user is interacting with is responsive. These kinds of issues were worked out in the 1960's and are covered in university level computer science clases such as "operating system design 101".
Mac OS X is very good at this. I can write a resource hog app and un it in the background but at the same time iTunes will play music without skipping. Even on an old G4 machine.
You also can't argue that MS has to suport all kinds of different hardware. Linux and BSD Unix support a much wider range of hardware.
"Panels on the moon would require vastly more infrastructure to support them"
I disagree. They could be even lighter on the moon. For eaxmple make them paper thin and then unroll them and leave them face up on the ground. There is no wind to blow dust onto them like on Mars. You need nearly zero support structure
But the down side it they could not rotate to track the sun so you need a much larger array nd a much larger battery system because of the larger period of darkness.
My idea would be to use both. Let the nuke plant run the base loads and then do other things like charging the rover's battery or driving robots trackers, running pumps and so on when the sun is up. That way you don't need the huge battery. The solar array would only be usfull about one wek a month but could be put to good use
Why not Solar.
What happens when it's dark? The moon rotates and it's dark for 14 days then light for 14 days. A solar power system would need batteries were large enough to last for 14 days. We are talking literaly tons of batteries that would have to be soft landed on tha moon and then assembled.
The solar array would have to be sized to about 4 times the desired output because the sun angle would moslty be non-optimal and it would have to use at last half the array's output to recharge the battery
Yes it could be done but it is worth doing a trade stdu to compare cost of each.
No, the ISS can't be the most expensive science project because it has little to do with science. ISS is a project alright but not a science project. Kind of an anti-science project realy as it's main purpose seems to be to suck up money from science related stuff.
"Anyway, Tennis is a relatively trivial example but things that happen in the physical world where physical forces are in play do not tolerate internet like latency very well. You cannot send xon/xoff like flow control signals to reality."
That is correct, so the way it would work is to isue higher level commands. Much like a coach would to a player. The coach gives only higher level statigy like "Stay more left of center and move up a bit." As robots become better they will need les and less real-time control.
"...a dot in the south pacific? if one were given to random chance, that's a lot of wasted souls in outrigger canoes in watery graves...."
People who lived on those islands tended to multiply and then run out of food. So they kill each other and took over the the limited land. The prehistory of Hawaii is one of endless wars and food and land rationing.
What happened is that people had to escape certain death by taking to the ocean in slim hope of finding another place. Over the centuries one in a thousand did find that other place. Most did not take to the oceans and were simply killed and their land taken.
Yes. I think that's the only answer that fits. Whoever first walked to North America must have come from North Asia. So it appears that maybe the population of N. Asia had a mix of peoples in it
Blu-Ray will be the LAST physical media. Samsung knows this and is not investing in R&D for the next media format. They've given that up and are now working on the new big dis[play technology.
"..in standard formats (e.g. CompactFlash, SD), you can just copy them bit for bit...."
No. The card card contain a active cryptographic device. But yes you could copy the data bit for bit but all you'd have is the encryted data. The point of using flash is that every copy is different because every SD card uses a different key and the key can't be read out of the card.
Don't say it's only a matter of time until it is cracked. Only poorly designed systems have been cracked the good ones have stood up for decades.
"...The chunk that actually broke off was 10 times the size predicted. Not sure why the big difference,..."
Some things are simply hard to predict. For example, we can predict that if a brick is thrown at a glass window the window will shatter but no one can predict the exact location of all the cracks in the glass in advance.
The falling leave problem is the same. We know it will fall down but can't predict the exact location.
All of this kind of stuff happens because very small, to small to measure initial condidtions cause huge changes in the result.
Recycling (except for aluminum cans and papers) uses more energy and costs more than creating new material
No. Recycling steel is very cost effective. Most low grade steel, like rebar is made from recycled material because it is cheaper.
Plastic bottles are used in big numbers buy both industry in Koera and the carpet manufactures in the American South.
Green waste is finding good use too. The price of mulch has falled which means more is being used which means there is less need for water, insectisides and so on.
ANd then the biggest use of re-cycling is that land fills are not being filled. In most places it costs about $50 a truckload to dump trash in a landfill. Not including the cost of the fuel to move the trash. So even if re-cycling were not 100% break even cost effective the over all saving is great when you factor in the saving in the cost to dump in a land fill. In other works even if you only get $1 per truckload of plastic soda bottles that is better than paying $100 per truckload to put them in a land fill.
Since there's evidence of multiple cycles of warming and cooling on the planet, another reason might be that cycling warming and cooling is a normal pattern for our planet.
Yes this is also happening. What we see the the supper imposed effects of both. There are several effects all gouing on at the same time. To sort it out you need lots of data and good statistical technique.
My opinion of this is that because coal and oil are the cheapest form of energy people will continue to dig it up and burn it until such time as coal and oil become so scarce and expensive that no sane person would burn them. You can't fight greed. But the good news is that as China and others come up to Western standards it's only 50 years until oil will become to valuable to burn. I think the problem is real but will self-correct itself inside of 75 years.
You don't have the shred the entire drive. I've seen security guys disassemble the drive and place each platter on a belt sander. Simply removing the coating is enough. Welding torches do a good job too. the coating come off with enough heat too.
It all depends what's on the drive in most cases the over write the data 10 times with random data is "good enough" but with other types of data physical destruction is the only option. There are many, may agencies each with their own rules about this.
Your ball of foil was NOT a faraday cage. Or rather it was a very poorly constructed one. You need a fully conductive enclosure with very tight seems. In professionally constructed enclosures you will notice things like copper gaskets and closely spaced bolts, doors where a copper knife edge is forced by a cam into a narrow slot. Even after the enclosure is built it must be tested and small air-gaps found and repaired.
In your case, a foil to foil joint is likely not conductive. Oxide coating on the Al foil acts as an insulator and you get a few ohms of resistance, at least.
Try another test using copper or brass pipe. Put the phone inside a length of pipe and screw end caps over each end of the pipe. Tighten until the bare metal on metal threads from a gas tight seal.
Yes this can be done. It is done. But in engineering we typically work from a set of requirements. This means we get a spec that says the part must fit in a given space or use only so much power or whtever. What you have is a solution looking for a requirement. In other words "I want to epoxy pot my M/B - Why would I do this"
If you must use this type of cooling why not simply us a nonconsecutive liquid? If you potted the M/B you'd have to pot all the conctors too. You never be able to replace a cable or RAn or move a jumper
The other thing that engineers get paid to do is come up with cost effective designs. In terms of cost this is just the wrong way to go.
The best way if you wont a compact and silent machine is to use the case itself as a heat sink. Conduct the heat to an aluminum case and let it radiate. Use a termally controlled fan as a backup for hot days and high loads. Oh wait isn't that what Apple does?
I don't buy the argument that Windows is targeted because it has 90% market share.
Do termites "target" wood frame building and avoid concrete structures because 90% of buildings are wood? Termite don't bother to each bricks because there are not so many bricks?
Same here. virues "eat" Windows and not any of the other OSes because Windows _can_ be eaten and the others can't. Windows is a soft target it was designed with the idea that it would run on an isolated desktop with no networking. Most all the other OSes are UNIX based. Unix was designed, back in the late 60's when computers were shared by many users. Unix was designed from the start to keep users isolated and unable to efect each other.
Yes Windows could improve but if it were to loose it's backwards compatabilty Microsoft would loose it's monopoly
What he's saying is that is appears to be written by some low-skill windows only programmers. That is the problem with open source. The whole world will see your work and judge you by it. Appearently this was written by some people who didn't know there was a world other then Windows on a PC.
I bet you can't read your doctor's prescription slips either. Every indutry has it's own set of terminology.
Why Chrome? If you are only viewing today's web pages then today's browsers are "good enough". But what Google hopes is that in the future you will be runing Google's on-line office suite, Google Maps and Google Mail. With Chrome all of these Google apps will run in their own tab, in their own thread and on one of the four or eight cores in your CPU.
Google knows full well that it takes many years to develope a mature web browser. This is their first step.
Today people don't run three or six different javascript aps all open at once. But Google, with Chrome hopes to change that.
This is related to the reason some people say "I don't need an 8 core machine". Of course that is true because you are running software that does not need 8 cores. But in 10 years your software will do different tasks. Maybe you will control the computer with hanf movments in front of a stereo 3D web cam while another camera watches where your eyes are looking, so as to update that part of the screen first. Same with browsers. Google is thinking ahead five to ten years.
"The show-stopper is(as of now) no NoScript/AdBlock! I've become spoiled with ad-free pages"
There are other ways to block ads. I run an HTML proxy server that also funtions to remove ads and other "junk". The proxy allows me to run any browser I want. If you have several computers the proxy works even better because it can block ads for an entire LAN and also cache common content.
orbital trajectories pretty well known, yes but hard to predict out to the future. So while you might know exactly where something is right now what you don't know is where exactly is will be next week. The unknowns are things like how much drag might slow it down, the atmosphere is not 100% constant and we don't know the shape of the object. Earth's gravity varies be location. We don't know about Solar activity and then there is the gravitational effects of the sun, moon and planets (Jupiter). With so many very small effects that can combine and interact there is a certain amount of error in any prediction
You have a room full of computers and you can afford $600?? Something is wrong.
Is the room full of cheap desktop PCs? if that is the case then your problem is that you yu are using to much power. Replace and consolidate the servers
Back to the "can't afford $600 problem: have you figured out what it will cost per month to run your $600 AC unit? I'm thinking that you can burn up $600 of power in 90 days. If you were to replace those servers you'd save a bundle. Look into something like one of Sun's "cool threads" servers. Typically one of these can replace a rack of PCs It's a very low power 16 core machine.
In the mean time it ooks like you've found your low cost solution, only that you are not only paying to much for power now but the $600 AC unit will double your power bill. Best to cut power usage with low power servers. Every wat that the server burns takes more than one wat to cool with an AC unit.