Wow, how is explaining the market dynamics of chip manufacturing a troll?!? Was it the use of the word capitalist? You do know that it's a perfectly valid economics term, right?
It's called binning and price discrimination. One is a technical/economic tool to maximize profits based on the non-perfect nature of chip manufacture, the other is the capitalists favorite tool to extract the maximum profit possible out of the consumer.
I know this is WAY late to the party, but Google has largely solved that problem with GEARS, you can work offline and wait for the cloud to come back to store your data. Works well for both the working on the airplane scenario as well as the "the system is down" scenario.
There are 3rd parties that sell the HP disk sleds for a LOT less than $250 per! Oh, and I'd like to find out why you don't trust BBW, tons of enterprises rely on them every day and I don't hear tons of horror stories about problems caused by the BBW.
If they're doing 48GB/node they better have speced that machine more than a year ago because everyone has known since this time last year that Nehalem is the best density solution and has the best $/MIP out there and almost all of the Nehalem boards support 72GB/node using cheap 4GB DIMM's. It really has been stupid to use anything else except possibly cell since the beginning of this year.
Uh, no enterprise class vendor just gives out pricing without working through a salesman or VAR. Mostly because they want the lead but also because they want to make sure the solution being quoted is the correct one for the customer. If you're serious it's not hard to get quotes, we had quotes for a half dozen different vendors and almost a dozen different solutions when we recently purchased a new SAN.
MTBF has NOTHING to do with a devices lifetime, it has to do with the average time for devices of that type to fail. If you have 200 of these running you should expect to have 1 to fail per year. This compares with the ~2% per year failure rate I've experienced in my datacenter over the last 3 years using mostly Segate enterprise class drives in a large number of different chassis from a variety of vendors. If you have thousands of these running you might have a failure per day, but still about one fourth the failures you would have had with traditional disks.
The problem is that such specialized sets are only furthering the problem of the helicopter parent generation, kids no longer understand how to participate in unstructured play. If everything is specialized and stylized into a specific role in a specific set then then how are kids supposed to come up with their own unique creations? Personally I've fought this trend by buying my son buckets of loose parts at garage sales, but at some point the huge piles of generic unstructured pieces will be gone, what then?
I'm talking about they layout as in printing setup. A good example is business cards, most people don't go through the couple hundred to thousand cards they typically get but the reason they give you so many is that the cost of printing the cards is dominated by the setup charges, it's cheaper to give everyone too many cards then it is to pay to setup even a small percentage of jobs for a second printing.
The concept of different ballot layouts doesn't compute here. There's one ballot. The candidates are in alphabetical order.
Uh, we don't do a national ballot in the US, each district/precinct has their own layout consisting of national, state, local, and precinct candidates and issues. I guess you could make it modular where the national and state parts are all printed in one large run (and perhaps they are) and then the local and precinct ballots are inserts for that general ballot but you still have to pay for the layout of those local ballots. Not only that but a large centralized economical print run just makes you less nimble so any ballot changes are a significant expense and might not be easily accommodated.
Way to not read the rest of my post, and btw I'm from Ohio which looks much like the majority of Canada. Paper's pretty much a commodity which means its price is set at an international level.
Uh, if you turn off write caching to the drive in Windows a crash means I/O stops when the OS stops sending bits to the I/O driver, not 30 seconds before or something like that. Modern SLC is reliable enough that you would have to be pushing Terrabytes per day for years to wear them out, if you are doing that then I think you will know how to disable this feature =)
I've always said Washington works best when they aren't working at all, I wonder what good an entire term without congresscritters would do for the country. (Remember, the opposite of CONgress is progress =)
Paper's NOT cheap (or more specifically printing isn't) which is one of the motivators behind electronic voting machines. Changes in wording or participants (death, withdraws due to scandal, etc) can mean reprinting all the ballots. You also need to print a ballot for every registered voter even if average turnout is well under 50%. You have all of the various precinct layouts so you have high setup costs for the print jobs. You also have to keep track of all that paper and move it around securely. I agree that it's the preferred system but it's far from perfect, much like Democracy it's the best system we have for the time being.
Actually Cuyahoga County (largest in Ohio) threw out Diebold after they had horrid technical issues in 2004. They didn't need any vote rigging to screw the majority Democrat vote here, the huge failure rate of the machines meant lines were long enough that people left in disgust. We went back to all paper registers and scantron style ballots and 2008 was MUCH smoother despite significantly higher turnout.
Actually wasn't it the Giza one that was destroyed when they launched it from the X-302 during the attack? Then the alternate stargate from the Russians was returned which was the one originally in Antarctica.
Uh, for the difference in price you buy and could build 3x the number of nodes needed and keep them powered off and still come out hundreds of thousands cheaper. In reality you might need a say 20% extra nodes and about the same in spare HDD's over the 5 year life of the system (any more than 5 years and it's probably not worth the power to keep them going). I have to question why they put the OS on a single HDD, flash would have been cheaper and more reliable. I also have to wonder WTF is up with using non-ES drives, the ES drives only cost a couple percent more and are actually built to run 24x7. Oh and anyone running a storage business with non-ECC ram is NOT someone I'm going to trust my data to!
Residential rates in NE Ohio are similar, it's 10.5c per kWh, but delivery, tax, cost recovery (stupid planned deregulation allowed the utility to recover the cost of their plants twice by now and we aren't going to get a deregulated market after the disaster in Illinois) makes it closer to 21c. Thankfully most of us have natural gas for heating or I'm not sure many people could afford to live around here (yeah gas spiked last year, but it was still half the cost per BTU of electric, this year it will be less than one fourth)
That's what I was thinking, but if maintenance is cheap enough it's not too bad. $70,000 per home supplied amortized over say a 50 year design life is $117/month which is on the low end of my monthly bill. Of course that ignores servicing debt and distribution so it's definitely more expensive then most current options but if you are a small island nation with lots of wealth spending 2x as much for electricity probably isn't a big deal compared to global warming wiping out half your landmass.
Power usage is actually going down per unit work, by a LOT due to virtualization. Compared to my standard server from just 3 years ago I can do 17:1 virtualization today without any major over-subscription of resources.
Try 32 half width/half height blades in 10U each running 2x6 core CPU's running VM's (actually I just checked and HP doesn't have a 2x blade based on these CPU's for some reason despite the fact that they are lower power than the 50W or 80W Intel's they currently use for the 2x). Of course the problem is that most VM implementations are memory not CPU bound so Intel wins for VM density with 192GB in a half height blade (HP BL460 G6). 512 cores and 12TB of ram in one rack is an amazing amount of computing power, but cooling that kind of power density is difficult to say the least.
Slashdot Mirror
Wow, how is explaining the market dynamics of chip manufacturing a troll?!? Was it the use of the word capitalist? You do know that it's a perfectly valid economics term, right?
It's called binning and price discrimination. One is a technical/economic tool to maximize profits based on the non-perfect nature of chip manufacture, the other is the capitalists favorite tool to extract the maximum profit possible out of the consumer.
I know this is WAY late to the party, but Google has largely solved that problem with GEARS, you can work offline and wait for the cloud to come back to store your data. Works well for both the working on the airplane scenario as well as the "the system is down" scenario.
There are 3rd parties that sell the HP disk sleds for a LOT less than $250 per! Oh, and I'd like to find out why you don't trust BBW, tons of enterprises rely on them every day and I don't hear tons of horror stories about problems caused by the BBW.
If they're doing 48GB/node they better have speced that machine more than a year ago because everyone has known since this time last year that Nehalem is the best density solution and has the best $/MIP out there and almost all of the Nehalem boards support 72GB/node using cheap 4GB DIMM's. It really has been stupid to use anything else except possibly cell since the beginning of this year.
Uh, no enterprise class vendor just gives out pricing without working through a salesman or VAR. Mostly because they want the lead but also because they want to make sure the solution being quoted is the correct one for the customer. If you're serious it's not hard to get quotes, we had quotes for a half dozen different vendors and almost a dozen different solutions when we recently purchased a new SAN.
MTBF has NOTHING to do with a devices lifetime, it has to do with the average time for devices of that type to fail. If you have 200 of these running you should expect to have 1 to fail per year. This compares with the ~2% per year failure rate I've experienced in my datacenter over the last 3 years using mostly Segate enterprise class drives in a large number of different chassis from a variety of vendors. If you have thousands of these running you might have a failure per day, but still about one fourth the failures you would have had with traditional disks.
The problem is that such specialized sets are only furthering the problem of the helicopter parent generation, kids no longer understand how to participate in unstructured play. If everything is specialized and stylized into a specific role in a specific set then then how are kids supposed to come up with their own unique creations? Personally I've fought this trend by buying my son buckets of loose parts at garage sales, but at some point the huge piles of generic unstructured pieces will be gone, what then?
I'm talking about they layout as in printing setup. A good example is business cards, most people don't go through the couple hundred to thousand cards they typically get but the reason they give you so many is that the cost of printing the cards is dominated by the setup charges, it's cheaper to give everyone too many cards then it is to pay to setup even a small percentage of jobs for a second printing.
The concept of different ballot layouts doesn't compute here. There's one ballot. The candidates are in alphabetical order.
Uh, we don't do a national ballot in the US, each district/precinct has their own layout consisting of national, state, local, and precinct candidates and issues. I guess you could make it modular where the national and state parts are all printed in one large run (and perhaps they are) and then the local and precinct ballots are inserts for that general ballot but you still have to pay for the layout of those local ballots. Not only that but a large centralized economical print run just makes you less nimble so any ballot changes are a significant expense and might not be easily accommodated.
Way to not read the rest of my post, and btw I'm from Ohio which looks much like the majority of Canada. Paper's pretty much a commodity which means its price is set at an international level.
Uh, if you turn off write caching to the drive in Windows a crash means I/O stops when the OS stops sending bits to the I/O driver, not 30 seconds before or something like that. Modern SLC is reliable enough that you would have to be pushing Terrabytes per day for years to wear them out, if you are doing that then I think you will know how to disable this feature =)
I've always said Washington works best when they aren't working at all, I wonder what good an entire term without congresscritters would do for the country. (Remember, the opposite of CONgress is progress =)
Paper's NOT cheap (or more specifically printing isn't) which is one of the motivators behind electronic voting machines. Changes in wording or participants (death, withdraws due to scandal, etc) can mean reprinting all the ballots. You also need to print a ballot for every registered voter even if average turnout is well under 50%. You have all of the various precinct layouts so you have high setup costs for the print jobs. You also have to keep track of all that paper and move it around securely. I agree that it's the preferred system but it's far from perfect, much like Democracy it's the best system we have for the time being.
Actually Cuyahoga County (largest in Ohio) threw out Diebold after they had horrid technical issues in 2004. They didn't need any vote rigging to screw the majority Democrat vote here, the huge failure rate of the machines meant lines were long enough that people left in disgust. We went back to all paper registers and scantron style ballots and 2008 was MUCH smoother despite significantly higher turnout.
Ah, I had forgotten about the switch with the Asgard ship.
Nighttime lows only get down to -4C (25F) so yeah basically t-shirt weather to a midwesterner.
Actually wasn't it the Giza one that was destroyed when they launched it from the X-302 during the attack? Then the alternate stargate from the Russians was returned which was the one originally in Antarctica.
Uh, for the difference in price you buy and could build 3x the number of nodes needed and keep them powered off and still come out hundreds of thousands cheaper. In reality you might need a say 20% extra nodes and about the same in spare HDD's over the 5 year life of the system (any more than 5 years and it's probably not worth the power to keep them going). I have to question why they put the OS on a single HDD, flash would have been cheaper and more reliable. I also have to wonder WTF is up with using non-ES drives, the ES drives only cost a couple percent more and are actually built to run 24x7. Oh and anyone running a storage business with non-ECC ram is NOT someone I'm going to trust my data to!
Not a lot more, my bill works out to about 21c per kWh so you're paying about 1/3rd more if that's the full bill rate and not the per-unit rate.
Residential rates in NE Ohio are similar, it's 10.5c per kWh, but delivery, tax, cost recovery (stupid planned deregulation allowed the utility to recover the cost of their plants twice by now and we aren't going to get a deregulated market after the disaster in Illinois) makes it closer to 21c. Thankfully most of us have natural gas for heating or I'm not sure many people could afford to live around here (yeah gas spiked last year, but it was still half the cost per BTU of electric, this year it will be less than one fourth)
If you are a small island nation with a large population land tends to be very scarce.
That's what I was thinking, but if maintenance is cheap enough it's not too bad. $70,000 per home supplied amortized over say a 50 year design life is $117/month which is on the low end of my monthly bill. Of course that ignores servicing debt and distribution so it's definitely more expensive then most current options but if you are a small island nation with lots of wealth spending 2x as much for electricity probably isn't a big deal compared to global warming wiping out half your landmass.
Power usage is actually going down per unit work, by a LOT due to virtualization. Compared to my standard server from just 3 years ago I can do 17:1 virtualization today without any major over-subscription of resources.
Try 32 half width/half height blades in 10U each running 2x6 core CPU's running VM's (actually I just checked and HP doesn't have a 2x blade based on these CPU's for some reason despite the fact that they are lower power than the 50W or 80W Intel's they currently use for the 2x). Of course the problem is that most VM implementations are memory not CPU bound so Intel wins for VM density with 192GB in a half height blade (HP BL460 G6). 512 cores and 12TB of ram in one rack is an amazing amount of computing power, but cooling that kind of power density is difficult to say the least.