Guessing they didn't do the geotechnical survey until the groundbreaking, but value engineering might also be at work. Until they hit a critical mass of launches it might be cheaper to use other facilities. So, decide to surcharge the soil for a couple years rather than piling and using thicker slabs, save lots of money.
Believe it or not, the sales folks have checks and balances. The legal department is one that should keep them in check on some items.
As for the question on how many servers it takes, there are answers. Ideally, the engineers get involved in the business aspects enough to understand that if then need $100k for servers today the project is dead, so what can be done to make it better. One common strategy is to scale and build redundancy over time, such that today we invest $30k, and for each of the next four years we do the same. (Yes, more money total, but the cash flow might be better.) Using "the cloud" as an alternate example, the engineer can decide to defer the capital investment until proof of concept and initial roll-out is complete and phase in new equipment.
I work with one non-profit to whom cashflow is everything. They make sub-optimal decisions because of it, but that is what they must do to meet their mission. I also work with a defense contractor guilty of the opposite: there is money left in the budget so all these wish list items need to be completed before the end of the year. I work with banks, who switch preferences between operating costs and capital costs based on management rotations.
Point being, either engineers need to be engaged in the business process, or the managers and/or sales people need to understand engineering. When the association is broken so is the business.
If you want to know how that works, look back at California's electric market deregulation in 2000/2001. Real time pricing is unfortunately very easy to game.
Over the last 5-10 years you have had a large number of power plants re-powered from oil to natural gas and from once-through cooling to cooling tower operation. I could be way off on my numbers, but I believe the cost is around $1MM/MWh typically, and generally amortized as a 30 year investment. So, in order to pay off those expenses, you are looking at whatever the existing (wholesale) cost of electricity is, and adding the cost of new renewable sources to it (roughly triple the wholesale cost amortized out). In addition, you need to add effective peaking capacity back in, which right now is sodium batteries.
The wholesale energy cost today is $0.03, give or take, so you would see energy costs go up by about $0.20/kWh to make it happen. In California this would be a retail cost of around $0.30/kWh compared to a blended $0.15 today.
Which seems kind of manageable. It would really suck for people with McMansions in Texas that currently have $3-400/summer month electricity bills seeing it go up to $6-800 per month... but market factors should push a solution.
The problem, which the electrical utilities are all painfully aware of, is that at such a point, what is their value-add? Would you see a mass exodus from the grid? (If so, would people just run a crappy little genset on days without adequate sun, making matters worse?) Whomever is left on the grid at that point is going to see costs closer to $1/kWh which is not viable in terms of investments that go out 30 years.
So, the alternative becomes letting the generators go bankrupt and/or bailing them out in order to get the costs off the books quickly. Then you need to bail out the consumers who made poor decisions relative to the previous utility assumptions. Oh, then you need to bail out cities, since they won't have enough land area for renewable resources to make them self-sufficient... so they are disproportionately impacted.
I think it can and should be done, but the horizon is likely closer to 20 years, and we need to get new nuclear plants permitted now and completed within 10 years. It is likely the only thing that would make financial sense. Less developed nations could be able to do it in an economically viable manner faster with the right technologies.
Your point can be boiled down to two things: engineers don't appreciate sales/marketing, and management doesn't anticipate the risks that are identified for them by engineers turning into real problems.
These failures rest on the engineers.
Yes, I am an engineer. I am also a manager and a piss-poor salesman. (Also a janitor and adult-daycare specialist.)
As an engineer, I must understand the challenges faced in the sales process. Namely, people often ask for things they can't have or that won't work. If that issue is identified in the pre-sales process, you won't get the work, but someone else will. When it is identified after thorough analysis then there is a different level of consideration, and a good engineer works within the process.
My favorite CxO that I have worked with over the years is a true genius. When I am asked to explain things to him, he insists that I keep it at a third grade level in my use of terms, process, and language. After I have presented, he proceeds to ask questions, generally using correct technical terms rather than whatever "third grade" terms I used, showing real insight into the issues, and more importantly into the business impact of the issues.
The challenge often comes down to making sure that the nuanced specificity of an engineer's response does not cloud the bigger issues. As an engineer I know there are very few black and white questions, but "it depends" is a non-productive answer.
There is a level of truth to this, but you need a base competency in-house to understand and champion the efforts.
Also, the general gripe of six-figure salaries in government isn't the base pay, it is all the benefits that are completely inconsistent with the private sector... while generally not being that much of a discount to private sector in salary.
You had me until the last paragraph... mostly. What is important is median family income and average family income statistics compared to the cost of living. In roughly a 5-mile radius of my home the median income is about $60-65k, while the average is $95-100k, and 10% have incomes over $200k and 20% are below the poverty line. Not a perfect measure, but gives an idea of the imbalance. Labor participation rates is a huge problem, and there are many different kinds of disenfranchised in that regard.
As for the last paragraph, "growth is the easiest way to hide incompetence." (--me). Immigration and population growth in general increase economic activity and allow for "progress" without focusing on just costs. With zero or negative growth, you suffer brain drain, lack of consumer spending, and are left with the government as employer of last resort. Which would you rather have?
Actually their design likely accommodates live load pretty well, since it has a bunch of columns relatively close together, and the pods have their own structural integrity. My guess is that the pods would be pretty close to 20T each. They could even have the columns taper outward for a larger base footprint effectively. If my math is anywhere close to correct they need less than 1m^2 at the base for the dead loads. They would also need about another square meter at each column for power conduits at 600V, or something slightly more realistic at 10-15kV.
It would be an interesting concept to actually try to make work; I am guessing the starting point is more of a parabolic tower shape, or just a giant single-story structure to support a 150MW backup turbine and other similar "accessory" functions...
Very true. I remember an architect at (I think) Helmut Jahn's office pulling out a psychrometric chart and scaring the shit out of a senior HVAC engineer.
That said, most architects have challenges with building systems beyond a buzzword level.
Architects design towers to be iconic. This is a cool rendering. Impractical, but cool. However, the elegant solution would need to be integrated into a parabolic tower as the GP pointed out. It would need to address all the functional issues associated with a data center in an elegant manner to be exceptional.
Modern data centers may take functionality too far-- there is room for elegance and finesse, but it must "follow function" to quote some guy. Core elements of the function are communications, power, cooling, reliability, security, and maintainability. Their solution hits on two of the six, but only elegantly in form.
Back in the day, most data centers were 3 stories, often with one expansion floor above and below. It maximized the utility of the mainframe bus and tag system, ensuring that the most equipment possible could be in direct contact with each other.
The next generation was largely two story facilities, stacking MEP infrastructure and the raised floor area. This better accommodated high(er) density solutions, as you had larger but shorter chilled water pipes, shorter electrical feeders, etc.
Then, things moved to single story facilities. This was all well before EYP's modular data center concept. Single story facilities made it easier to use free cooling, swamp cooling, and other air-side economizer strategies. It improved construction time, as the structure was much more limited, and it made it a little easier to have absurd raised floor heights.
Local factors also played a role-- land costs, floor area needs, network issues, etc., so you still see a few multi-story facilities being built.
This concept is just another architect's penis, and apparently one that doesn't understand much about data centers. Sure you gain stack effect, at a cost of a whole lot of redundant maintenance efforts and absurd infrastructure components. A quick back of matchbox calculation is that each module would need at least 300kW critical (plus about 40kw parasitic load) to border on viable... at which point it would still be cheaper and more efficient to provide a containerized solution in a warehouse with a big old parabolic cooling tower for heat rejection.
Tend to agree; government agencies tend to have one exceptional person near the top and a whole lot of "others" filling the ranks. There are so many bad measures to control costs pushed onto contractors that it drives up compliance costs to insane levels. One person in my office has a full time job just dealing with invoices.
Not all government organizations are the same though. Generally it is when one gets a windfall of cash to spend that everything goes out the window. The $trillions of major jnrastructure upgrades will fall into this trap: it is cheaper to invest consistently over time than defer work and let it ballon up.
I really like it, although there is room for improvement. The app screen is awful, too easy to hit the wrong icon. Health and fitness functions are nice, love the "modular" watch face showing temperature, date, stock ticker, and activity icon-- it puts everything in such an easy place to access. Most used apps are the wallet for airline boarding passes, Apple Pay, weather, remote, map, phone, text/emoji, and oddly calendar.
Do miss my Welder watch though, and wish I could swim with it.
As for sales, my wife was noticing how many people had one on yesterday; I think she counted a couple dozen in the food court and 10 people out of 60 in another spot. They seem to be doing well, but I am sure there is some selection bias going on.
200-mile range is pretty damn good. On a real road trip, that means a stop every 3-4 hours. If you really are a 1,000 mile a day driver then rent a car for the trip! If you are fine with 600 miles per day, and two long stops en-route, it isn't that big of a deal.
Tesla needs to continue to offer multiple models for just that reason. If they can go from 50->80->130->210->350->500k vehicles per year within their current factory and with available suppliers then it isn't that hard to pull off. They basically need about a 4-5% increase in production capacity per month.
Comparing to narrow body airplane production, it would be an impossible task; certification and limited vendors for certain parts (currently seats) make faster ramp-ups impossible. However, the physical factory previously produced 300k vehicles per year, so getting to that point should be reasonably achievable.
Slashdot Mirror
A380 only needs ~80k lbf; while the engine would physically fit it must be significantly over-dimensioned.
I type if I want to be able to organize thoughts and actually read them later.
I hand write to form visual associations to remember things.
Guessing they didn't do the geotechnical survey until the groundbreaking, but value engineering might also be at work. Until they hit a critical mass of launches it might be cheaper to use other facilities. So, decide to surcharge the soil for a couple years rather than piling and using thicker slabs, save lots of money.
Leaves far too much room for McShitface.
:)
Can't make myself write $1M; it is like $1k... Nonsensical (financial) units...
Believe it or not, the sales folks have checks and balances. The legal department is one that should keep them in check on some items.
As for the question on how many servers it takes, there are answers. Ideally, the engineers get involved in the business aspects enough to understand that if then need $100k for servers today the project is dead, so what can be done to make it better. One common strategy is to scale and build redundancy over time, such that today we invest $30k, and for each of the next four years we do the same. (Yes, more money total, but the cash flow might be better.) Using "the cloud" as an alternate example, the engineer can decide to defer the capital investment until proof of concept and initial roll-out is complete and phase in new equipment.
I work with one non-profit to whom cashflow is everything. They make sub-optimal decisions because of it, but that is what they must do to meet their mission. I also work with a defense contractor guilty of the opposite: there is money left in the budget so all these wish list items need to be completed before the end of the year. I work with banks, who switch preferences between operating costs and capital costs based on management rotations.
Point being, either engineers need to be engaged in the business process, or the managers and/or sales people need to understand engineering. When the association is broken so is the business.
If you want to know how that works, look back at California's electric market deregulation in 2000/2001. Real time pricing is unfortunately very easy to game.
Over the last 5-10 years you have had a large number of power plants re-powered from oil to natural gas and from once-through cooling to cooling tower operation. I could be way off on my numbers, but I believe the cost is around $1MM/MWh typically, and generally amortized as a 30 year investment. So, in order to pay off those expenses, you are looking at whatever the existing (wholesale) cost of electricity is, and adding the cost of new renewable sources to it (roughly triple the wholesale cost amortized out). In addition, you need to add effective peaking capacity back in, which right now is sodium batteries.
The wholesale energy cost today is $0.03, give or take, so you would see energy costs go up by about $0.20/kWh to make it happen. In California this would be a retail cost of around $0.30/kWh compared to a blended $0.15 today.
Which seems kind of manageable. It would really suck for people with McMansions in Texas that currently have $3-400/summer month electricity bills seeing it go up to $6-800 per month... but market factors should push a solution.
The problem, which the electrical utilities are all painfully aware of, is that at such a point, what is their value-add? Would you see a mass exodus from the grid? (If so, would people just run a crappy little genset on days without adequate sun, making matters worse?) Whomever is left on the grid at that point is going to see costs closer to $1/kWh which is not viable in terms of investments that go out 30 years.
So, the alternative becomes letting the generators go bankrupt and/or bailing them out in order to get the costs off the books quickly. Then you need to bail out the consumers who made poor decisions relative to the previous utility assumptions. Oh, then you need to bail out cities, since they won't have enough land area for renewable resources to make them self-sufficient... so they are disproportionately impacted.
I think it can and should be done, but the horizon is likely closer to 20 years, and we need to get new nuclear plants permitted now and completed within 10 years. It is likely the only thing that would make financial sense. Less developed nations could be able to do it in an economically viable manner faster with the right technologies.
Oooh, and past and present senators too!
Last time i wrote her she basically replied with a form letter saying I was wrong. Good luck...
No, sadly Carly was the most viable out of the lot. She might have even gotten 10% of the vote!
Have you seen what passes for a republican candidate in California? The Governator is about the only republican that could replace her easily.
Your point can be boiled down to two things: engineers don't appreciate sales/marketing, and management doesn't anticipate the risks that are identified for them by engineers turning into real problems.
These failures rest on the engineers.
Yes, I am an engineer. I am also a manager and a piss-poor salesman. (Also a janitor and adult-daycare specialist.)
As an engineer, I must understand the challenges faced in the sales process. Namely, people often ask for things they can't have or that won't work. If that issue is identified in the pre-sales process, you won't get the work, but someone else will. When it is identified after thorough analysis then there is a different level of consideration, and a good engineer works within the process.
My favorite CxO that I have worked with over the years is a true genius. When I am asked to explain things to him, he insists that I keep it at a third grade level in my use of terms, process, and language. After I have presented, he proceeds to ask questions, generally using correct technical terms rather than whatever "third grade" terms I used, showing real insight into the issues, and more importantly into the business impact of the issues.
The challenge often comes down to making sure that the nuanced specificity of an engineer's response does not cloud the bigger issues. As an engineer I know there are very few black and white questions, but "it depends" is a non-productive answer.
There is a level of truth to this, but you need a base competency in-house to understand and champion the efforts.
Also, the general gripe of six-figure salaries in government isn't the base pay, it is all the benefits that are completely inconsistent with the private sector... while generally not being that much of a discount to private sector in salary.
You had me until the last paragraph... mostly. What is important is median family income and average family income statistics compared to the cost of living. In roughly a 5-mile radius of my home the median income is about $60-65k, while the average is $95-100k, and 10% have incomes over $200k and 20% are below the poverty line. Not a perfect measure, but gives an idea of the imbalance. Labor participation rates is a huge problem, and there are many different kinds of disenfranchised in that regard.
As for the last paragraph, "growth is the easiest way to hide incompetence." (--me). Immigration and population growth in general increase economic activity and allow for "progress" without focusing on just costs. With zero or negative growth, you suffer brain drain, lack of consumer spending, and are left with the government as employer of last resort. Which would you rather have?
Actually their design likely accommodates live load pretty well, since it has a bunch of columns relatively close together, and the pods have their own structural integrity. My guess is that the pods would be pretty close to 20T each. They could even have the columns taper outward for a larger base footprint effectively. If my math is anywhere close to correct they need less than 1m^2 at the base for the dead loads. They would also need about another square meter at each column for power conduits at 600V, or something slightly more realistic at 10-15kV.
It would be an interesting concept to actually try to make work; I am guessing the starting point is more of a parabolic tower shape, or just a giant single-story structure to support a 150MW backup turbine and other similar "accessory" functions...
Very true. I remember an architect at (I think) Helmut Jahn's office pulling out a psychrometric chart and scaring the shit out of a senior HVAC engineer.
That said, most architects have challenges with building systems beyond a buzzword level.
Architects design towers to be iconic. This is a cool rendering. Impractical, but cool. However, the elegant solution would need to be integrated into a parabolic tower as the GP pointed out. It would need to address all the functional issues associated with a data center in an elegant manner to be exceptional.
Modern data centers may take functionality too far-- there is room for elegance and finesse, but it must "follow function" to quote some guy. Core elements of the function are communications, power, cooling, reliability, security, and maintainability. Their solution hits on two of the six, but only elegantly in form.
Back in the day, most data centers were 3 stories, often with one expansion floor above and below. It maximized the utility of the mainframe bus and tag system, ensuring that the most equipment possible could be in direct contact with each other.
The next generation was largely two story facilities, stacking MEP infrastructure and the raised floor area. This better accommodated high(er) density solutions, as you had larger but shorter chilled water pipes, shorter electrical feeders, etc.
Then, things moved to single story facilities. This was all well before EYP's modular data center concept. Single story facilities made it easier to use free cooling, swamp cooling, and other air-side economizer strategies. It improved construction time, as the structure was much more limited, and it made it a little easier to have absurd raised floor heights.
Local factors also played a role-- land costs, floor area needs, network issues, etc., so you still see a few multi-story facilities being built.
This concept is just another architect's penis, and apparently one that doesn't understand much about data centers. Sure you gain stack effect, at a cost of a whole lot of redundant maintenance efforts and absurd infrastructure components. A quick back of matchbox calculation is that each module would need at least 300kW critical (plus about 40kw parasitic load) to border on viable... at which point it would still be cheaper and more efficient to provide a containerized solution in a warehouse with a big old parabolic cooling tower for heat rejection.
Tend to agree; government agencies tend to have one exceptional person near the top and a whole lot of "others" filling the ranks. There are so many bad measures to control costs pushed onto contractors that it drives up compliance costs to insane levels. One person in my office has a full time job just dealing with invoices.
Not all government organizations are the same though. Generally it is when one gets a windfall of cash to spend that everything goes out the window. The $trillions of major jnrastructure upgrades will fall into this trap: it is cheaper to invest consistently over time than defer work and let it ballon up.
I really like it, although there is room for improvement. The app screen is awful, too easy to hit the wrong icon. Health and fitness functions are nice, love the "modular" watch face showing temperature, date, stock ticker, and activity icon-- it puts everything in such an easy place to access. Most used apps are the wallet for airline boarding passes, Apple Pay, weather, remote, map, phone, text/emoji, and oddly calendar.
Do miss my Welder watch though, and wish I could swim with it.
As for sales, my wife was noticing how many people had one on yesterday; I think she counted a couple dozen in the food court and 10 people out of 60 in another spot. They seem to be doing well, but I am sure there is some selection bias going on.
FWIW, a yubikey isn't that hard to set up for 2FA on the PC.
200-mile range is pretty damn good. On a real road trip, that means a stop every 3-4 hours. If you really are a 1,000 mile a day driver then rent a car for the trip! If you are fine with 600 miles per day, and two long stops en-route, it isn't that big of a deal.
Tesla needs to continue to offer multiple models for just that reason. If they can go from 50->80->130->210->350->500k vehicles per year within their current factory and with available suppliers then it isn't that hard to pull off. They basically need about a 4-5% increase in production capacity per month.
Comparing to narrow body airplane production, it would be an impossible task; certification and limited vendors for certain parts (currently seats) make faster ramp-ups impossible. However, the physical factory previously produced 300k vehicles per year, so getting to that point should be reasonably achievable.
No, (digitally) signed authorizations should be provided.
For sterile products or drugs?
Who really cares about sterile products...