It doesn't really matter how many people are employed there; it would likely be at least 50-100 per shift running three shifts. The value is that you get a low-margin part of the business subsidized and you allow for better risk management. Batteries are pretty heavy, and if you can build a factory in the midwest then they will favor domestic consumption.
The biggest problem with a battery factory is dealing with the EPA (for good reason).
The DOD budget is about 25% of the federal budget. Over half of the budget is medicare/medicaid/social security. Nobody talks about the elephant in the room...
The Buffalo Terastation uses a software RAID, which slows it considerably, with the side benefit of being nearly impossible to recover if it crashes.
It does support SMB, NFS, and AFS out of the box though.
These boxes are cheap crap, and have a very limited useful lifespan. Our company lost a good deal of information when ours crapped out after 366 days. (Yes, we had backups, No they weren't perfect. They happened to be with me halfway around the globe at the time...)
Really seems like the product offerings in this space are limited usability, poor reliability, imperfect implementations, and grossly overpriced. Doing it over again, I would go for a build-it-yourself box hands down.
I think in time, Google will have a vested interest in becoming an ISP. While distributed data centers might reduce the total long-haul bandwidth they need, getting preferred access to the local loops is what is critical for Google Apps and SaaS in general.
A Google or MS can quickly subvert the stranglehold the LECs have today in major metropolitan areas if they have a competitive incentive to do so.
You could argue that the netbooks are the best users for something like this; the heat dissipation options with air are much more limiting in compact designs than larger machines. I know the venting on my AA1 is ugly when compared to the macs, and the noise level is pretty significant. Something with a very small membrane pump could do wonders.
Only problem is that a pump failure will immediately toast the system, and you have to spend real money to keep the piping connections from leaking.
California spends ~$7k per student, which assuming a 20-person class size, 5 classes per teacher in a 6 period day yields $116k per "teacher." If 50% of that per teacher budget is direct salary then you pay teachers $58k on average.
That 50% number is consistent with the service industry. (1/3 direct salary, 1/3 direct and indirect overhead, and 1/3 profit). The salary is what I would think is a reasonable wage for a teacher (on a state average), although by no means excessive. $8k/pupil is likely more fair, which would be $66k average per teacher.
Backing out other numbers, $7k allows for a construction cost of a school of around $200/SF with good maintenance programs (and 4% bond interest), a 1:130 ratio of administrators to students, and $350-500 per student in teaching supplies. All of this requires a well-run operation, but isn't rocket science.
It is hard for me as a business person to understand where more money is going to help the problem. I really don't want to force the schools to benchmark capital spending, administrative overhead, direct and indirect labor overhead and all that other crap so I can feel good about the money they are getting and the salaries teachers are paid.
While I went to a great public school in my younger years (which became great by being very bad decades earlier), I hate the argument that schools can fix themselves.
Parents are the first problem. When parents don't encourage their children to excel in school, work hard, and help them learn, they are destine to fail. Unfortunately, it only takes a few bad parents to create a bad class and a few bad classes to make a bad school.
The next problem is apathy. If kids don't think any of their actions can improve their own success, they won't succeed. Economic development is one way to reduce apathy by creating opportunities.
The ultimate question is what can be done to make a bad school good again. If the only answer is closing the doors or just turning it into a youth detention facility, then vouchers are a great idea. It's just kind of a shitty way to let people drop through the cracks.
Why can't ISPs cut off their end users that try to contact any of the control domains? Easier to isolate the bots than kill the head; they need to find a new attack vector to re-infect the machines which will take longer than reestablishing the command and control networks... assuming the end-users are forced to install appropriate updates before being allowed back online.
It is much easier to use VGA for home theater and conference rooms than DVI (just need a balun on either end and you can use CAT-5 structured cabling with VGA). VGA supports the same resolutions as DVI and HDMI. This alone is a huge reason why the whole HDCP issue is a pain in the ass.
It is a battle that we, the customers, should not have to put up with.
DC Power for data centers is a good technology, but if you are suggesting 48VDC you must own copper futures! Most of the data center scale systems proposed today are 600VDC (nominally), as you can easily get to 6MW on a single bus. 48VDC would only give you 300kW, assuming your runs were very short. (Bus size based on a 4-wire, 5,000A bus duct with two phase busses per pole.)
A heat wheel is an odd way to cool a data center; it requires fairly low outside air temperatures to work. We've had better luck with indirect evaprative coolers pre-cooling the hot-aisle return air and discharging to the space. Unless your wet-bulb temperature is below 60F, you will still need some compressor cooling, but since wet bulb temperature is never higher than dry bulb you get much better range and efficiency.
I was actually working with a client on an SLA with a co-lo facility recently. The co-lo would only commit to three nines, despite the fact that their infrastructure should theoretically support a solid four nines. They keep that extra order of magnitude in their back pocket-- be it to improve profit margins later or "value-added" services soon to come.
You have to keep in mind that statistical reliability and real-world availability are different animals. Some of it comes down to luck, and some is about looking at the full life-cycle.
My company's IT budget is likely a good order of magnitude smaller-- 20 person organization. Not having your network on UPS is just stupid!
We are in a major metropolitan area, but we have a UPS for workstations even if they are being used by a part-time student just above minimum wage. We get about four hits a year, and that alone is enough for it to make sense.
We also have a T1 and ADSL from different providers. While automated failover isn't in place, it is on our list as time allows.
The concept of "unplanned downtime" seems to originate in the banking world, where something as benign as daylight savings time could force you to take down the mainframe for two hours. It has unfortunately spread to other industries (healthcare records management pops up). The real question is if Google's application architecture requires planned downtime for the service as a whole or individual users.
Based on their roots, I would expect them to be able to do any upgrades in the ten minute window they exclude from their SLA.
CAD:USD exchange rate isn't as favorable as it used to be. They priced aggressively before, and with the rate change they have to re-price. Won't be good for ROW sales in a recession...
With the US power system, you do avoid four high loss (DC:AC or AC:AC) power conversions and replace them with two lower loss (DC:DC) conversions. But, compared to the ROW electrical systems, you only save two AC:AC conversions which will just gain you two points or so.
I like 600VDC as a solution, but it will only work well for the biggest consumers where you can justify a significant increase capital cost with the energy savings. It's nice to have a single 4.8MW critical power bus (with a couple spares)-- makes load management much easier and reduces space requirements significantly.
I actually do work in a data center that has had air-side economizers since 1985. They have pretty good statistics on equipment failure, independent of ambient temperature and humidity. Seven months isn't an issue, but when you get towards 2-3 years failure rates jump way up. Pollution is the biggest cause, but dust is also a problem over time.
Well put. Key to financial success is owning your own business and living on less than minimum wage for a year or two in the hopes of making twice as much afterwards. If you can moonlight to get things started, it isn't too bad of a hit...
Of course, it adds all kinds of other stresses to your life, and you spend much more energy doing things like invoicing, financing, dealing with insurance...
If you are in an underserved area without competition, what does it take in practical and financial terms to build out a community ISP that can compete with the Comcasts of the world?
A 24-port ADSL2+ DSLAM and associated CPE would run about $200-300 per user, "last quarter mile" wiring could be anywhere from $300-3,000 per user.
What would you have to pay to get an OC3 for upstream connection?
I looked at a few years back when end-user bandwidth wasn't as much of an issue and you could get much better reach, but it still came back with a per-subscriber cost around $60/month. How big do you have to get to provide a better value than a Comcast?
Um, Google has quite a bit of cash on hand that can be used for just this type of capital programs. Arguably, it also does the most for the economy, since they actually spend money they have been hoarding, which creates jobs.
Slashdot Mirror
It doesn't really matter how many people are employed there; it would likely be at least 50-100 per shift running three shifts. The value is that you get a low-margin part of the business subsidized and you allow for better risk management. Batteries are pretty heavy, and if you can build a factory in the midwest then they will favor domestic consumption.
The biggest problem with a battery factory is dealing with the EPA (for good reason).
The DOD budget is about 25% of the federal budget. Over half of the budget is medicare/medicaid/social security. Nobody talks about the elephant in the room...
The Buffalo Terastation uses a software RAID, which slows it considerably, with the side benefit of being nearly impossible to recover if it crashes.
It does support SMB, NFS, and AFS out of the box though.
These boxes are cheap crap, and have a very limited useful lifespan. Our company lost a good deal of information when ours crapped out after 366 days. (Yes, we had backups, No they weren't perfect. They happened to be with me halfway around the globe at the time...)
Really seems like the product offerings in this space are limited usability, poor reliability, imperfect implementations, and grossly overpriced. Doing it over again, I would go for a build-it-yourself box hands down.
While the user page is the worst, I really hate the lack of a tree style expansion like we had with the slashdotter extensions in FF
I think in time, Google will have a vested interest in becoming an ISP. While distributed data centers might reduce the total long-haul bandwidth they need, getting preferred access to the local loops is what is critical for Google Apps and SaaS in general.
A Google or MS can quickly subvert the stranglehold the LECs have today in major metropolitan areas if they have a competitive incentive to do so.
GPS is great in places where streets really don't have names...
You could argue that the netbooks are the best users for something like this; the heat dissipation options with air are much more limiting in compact designs than larger machines. I know the venting on my AA1 is ugly when compared to the macs, and the noise level is pretty significant. Something with a very small membrane pump could do wonders.
Only problem is that a pump failure will immediately toast the system, and you have to spend real money to keep the piping connections from leaking.
California spends ~$7k per student, which assuming a 20-person class size, 5 classes per teacher in a 6 period day yields $116k per "teacher." If 50% of that per teacher budget is direct salary then you pay teachers $58k on average.
That 50% number is consistent with the service industry. (1/3 direct salary, 1/3 direct and indirect overhead, and 1/3 profit). The salary is what I would think is a reasonable wage for a teacher (on a state average), although by no means excessive. $8k/pupil is likely more fair, which would be $66k average per teacher.
Backing out other numbers, $7k allows for a construction cost of a school of around $200/SF with good maintenance programs (and 4% bond interest), a 1:130 ratio of administrators to students, and $350-500 per student in teaching supplies. All of this requires a well-run operation, but isn't rocket science.
It is hard for me as a business person to understand where more money is going to help the problem. I really don't want to force the schools to benchmark capital spending, administrative overhead, direct and indirect labor overhead and all that other crap so I can feel good about the money they are getting and the salaries teachers are paid.
While I went to a great public school in my younger years (which became great by being very bad decades earlier), I hate the argument that schools can fix themselves.
Parents are the first problem. When parents don't encourage their children to excel in school, work hard, and help them learn, they are destine to fail. Unfortunately, it only takes a few bad parents to create a bad class and a few bad classes to make a bad school.
The next problem is apathy. If kids don't think any of their actions can improve their own success, they won't succeed. Economic development is one way to reduce apathy by creating opportunities.
The ultimate question is what can be done to make a bad school good again. If the only answer is closing the doors or just turning it into a youth detention facility, then vouchers are a great idea. It's just kind of a shitty way to let people drop through the cracks.
Why can't ISPs cut off their end users that try to contact any of the control domains? Easier to isolate the bots than kill the head; they need to find a new attack vector to re-infect the machines which will take longer than reestablishing the command and control networks... assuming the end-users are forced to install appropriate updates before being allowed back online.
It is much easier to use VGA for home theater and conference rooms than DVI (just need a balun on either end and you can use CAT-5 structured cabling with VGA). VGA supports the same resolutions as DVI and HDMI. This alone is a huge reason why the whole HDCP issue is a pain in the ass.
It is a battle that we, the customers, should not have to put up with.
DC Power for data centers is a good technology, but if you are suggesting 48VDC you must own copper futures! Most of the data center scale systems proposed today are 600VDC (nominally), as you can easily get to 6MW on a single bus. 48VDC would only give you 300kW, assuming your runs were very short. (Bus size based on a 4-wire, 5,000A bus duct with two phase busses per pole.)
A heat wheel is an odd way to cool a data center; it requires fairly low outside air temperatures to work. We've had better luck with indirect evaprative coolers pre-cooling the hot-aisle return air and discharging to the space. Unless your wet-bulb temperature is below 60F, you will still need some compressor cooling, but since wet bulb temperature is never higher than dry bulb you get much better range and efficiency.
So they can audit total number of voters for each machine.
I was actually working with a client on an SLA with a co-lo facility recently. The co-lo would only commit to three nines, despite the fact that their infrastructure should theoretically support a solid four nines. They keep that extra order of magnitude in their back pocket-- be it to improve profit margins later or "value-added" services soon to come.
You have to keep in mind that statistical reliability and real-world availability are different animals. Some of it comes down to luck, and some is about looking at the full life-cycle.
My company's IT budget is likely a good order of magnitude smaller-- 20 person organization. Not having your network on UPS is just stupid!
We are in a major metropolitan area, but we have a UPS for workstations even if they are being used by a part-time student just above minimum wage. We get about four hits a year, and that alone is enough for it to make sense.
We also have a T1 and ADSL from different providers. While automated failover isn't in place, it is on our list as time allows.
Google appliance unfortunately is just for search. Here's to hoping they add app support as well in the future.
The concept of "unplanned downtime" seems to originate in the banking world, where something as benign as daylight savings time could force you to take down the mainframe for two hours. It has unfortunately spread to other industries (healthcare records management pops up). The real question is if Google's application architecture requires planned downtime for the service as a whole or individual users.
Based on their roots, I would expect them to be able to do any upgrades in the ten minute window they exclude from their SLA.
I think the real problem is that the scale of small business data storage will quickly outpace the capability of RAID5.
But the real issue to the summary is that a 6+1 array is less reliable than a single drive. True for almost anything, including the UPS.
CAD:USD exchange rate isn't as favorable as it used to be. They priced aggressively before, and with the rate change they have to re-price. Won't be good for ROW sales in a recession...
With the US power system, you do avoid four high loss (DC:AC or AC:AC) power conversions and replace them with two lower loss (DC:DC) conversions. But, compared to the ROW electrical systems, you only save two AC:AC conversions which will just gain you two points or so.
I like 600VDC as a solution, but it will only work well for the biggest consumers where you can justify a significant increase capital cost with the energy savings. It's nice to have a single 4.8MW critical power bus (with a couple spares)-- makes load management much easier and reduces space requirements significantly.
I actually do work in a data center that has had air-side economizers since 1985. They have pretty good statistics on equipment failure, independent of ambient temperature and humidity. Seven months isn't an issue, but when you get towards 2-3 years failure rates jump way up. Pollution is the biggest cause, but dust is also a problem over time.
Well put. Key to financial success is owning your own business and living on less than minimum wage for a year or two in the hopes of making twice as much afterwards. If you can moonlight to get things started, it isn't too bad of a hit...
Of course, it adds all kinds of other stresses to your life, and you spend much more energy doing things like invoicing, financing, dealing with insurance...
If you are in an underserved area without competition, what does it take in practical and financial terms to build out a community ISP that can compete with the Comcasts of the world?
A 24-port ADSL2+ DSLAM and associated CPE would run about $200-300 per user, "last quarter mile" wiring could be anywhere from $300-3,000 per user.
What would you have to pay to get an OC3 for upstream connection?
I looked at a few years back when end-user bandwidth wasn't as much of an issue and you could get much better reach, but it still came back with a per-subscriber cost around $60/month. How big do you have to get to provide a better value than a Comcast?
Um, Google has quite a bit of cash on hand that can be used for just this type of capital programs. Arguably, it also does the most for the economy, since they actually spend money they have been hoarding, which creates jobs.