The reality is they don't have the customer base to do modern processor development. Look for the next generation processors to get pushed into the future until being cancelled, ultimately.
Where I work we used Sun because of performance in the beginning, then because Solaris was superior to Windows. With the advent of Linux, the only reason we've kept them around is 64-bit address space. I really don't see what they offer over a server-class Athlon-64 running Linux. Except a price premium.
As far as services are concerned, they really put a big hole in their own foot. The multiplatform nature of Java prevents them from keeping a vendor-lock on customers the way IBM has with its mainframes. We can trasition any recent project to other hardware at any time.
No. Satellites are expensive because they are almost always mission critical hardware for their function. That dictates the extensive testing, expensive parts, etc. The repeaters on undersea cables cost almost three times a pound as much as current satellites, even though the cost of placing them is far lower per pound. Why? Because the loss of a repeater means the loss of millions of dollars in revenue until it can be raised and replaced or repaired.
Er... wrong. The extra reliability requirements of the satellite are only related to outage costs in that they determine how many spares you need. At current launch prices you need lots of reliability since each spare is very expensive to loft.
I'm not familiar with the undersea cable business, but I'm guessing the cost of the repeater is peanuts compared to the overall project cost. It would make sense, under those conditions, to optimize for reliability vs. cost, since redundancy means adding a whole extra cable.
Satellites are a different matter. Since the satellite is the entire system (as opposed to a piece of the whole), it makes more and more sense to optimize for cost as launch costs come down since you can boost your reliability with spares.
The cost of launching a GEO bird could drop to $10/lb next Monday, and a commsat would still cost tens or hundreds of millions of dollars, because the cost of an outage because of a failed bird remains the same
Today's commercial and military systems already have spares. Launch costs will never come down to $10/pound, but if they did you could design to much less stringent criteria, because each satellite could 1) be heavier, 2) have a shorter lifespan, and 3) be less reliable as a result of redundancy. Look, what makes these things expensive is they must be as reliable as possible within the limits of our technology without weighing very much. You're paying ten times the cost for that extra digit.
And it would still take weeks to months to replace the bird. They aren't built on an assembly line and never will be, there simply isn't a need for that many, nor is their space in GEO for them.
It wouldn't take weeks or months, as I said, since you would have another one ready to power up. And while it's true you wouldn't build them on an assembly line, your design costs don't change if you build one, ten, or twenty. The second one is much cheaper than the first. In any event the primary reason we don't "need" many is because they cost too much. Lower the price point and you get more customers as the service enters their price range.
As far as space in GEO, that's nonsense. What's in short supply is space for active satellites, since they can't be so close they are indistinguishable to ground-based receivers. But there's plenty of space for "dark" birds. In any case you can use directional antennae to "multiplex" the slot, and for other comm applications lower orbits make sense anyway - with cheap phased array antennae GEO isn't as necessary as it once was, and it has its (latency) drawbacks. ~
What a lot of people are missing here is the "virtuous spiral" of the cost.
When launch costs are lower you don't need satellites that last 15 years and tested to the nines. If I use the shuttle to launch it costs half a billion dollars to get my bird up (although a lot of that is picked up by the taxpayers), so I need something that's guarenteed to work. That means lots of expensive parts, and lots of expensive testing.
If I can launch cheaply I can afford to make cheaper satellites, since the cost of failure is lower. So now I need one less decimal place in my reliability, which means one less decimal place in the price. And I don't need the darned thing to work forever - a five year life might make more sense if I can replace it cheaply.
This makes the number of launches go up. Which makes the cost of the launch go down. Which makes the price of satellites go down. Take this loop a couple of times and you'll get closer to the actual production cost of the rocket, which is very low, in the grand scheme of things.
Slashdot Mirror
Except that the whole drawback to PDEs is... wait for it... THEY'RE TOO BLOODY NOISY!
SCO doesn't have enough money to stave off IBM's legal grinding wheel, since it can't raise more capital.
Baystar loses its corporate ass, which will make it harder for any company to use lawsuits as a business plan.
"I have nurtured you and made you strong. Attack my enemies forthwith and destroy all rivals to my power!"
Shouldn't mix this cough medicine with caffeine...
Where I work we used Sun because of performance in the beginning, then because Solaris was superior to Windows. With the advent of Linux, the only reason we've kept them around is 64-bit address space. I really don't see what they offer over a server-class Athlon-64 running Linux. Except a price premium.
As far as services are concerned, they really put a big hole in their own foot. The multiplatform nature of Java prevents them from keeping a vendor-lock on customers the way IBM has with its mainframes. We can trasition any recent project to other hardware at any time.
Er... wrong. The extra reliability requirements of the satellite are only related to outage costs in that they determine how many spares you need. At current launch prices you need lots of reliability since each spare is very expensive to loft.
I'm not familiar with the undersea cable business, but I'm guessing the cost of the repeater is peanuts compared to the overall project cost. It would make sense, under those conditions, to optimize for reliability vs. cost, since redundancy means adding a whole extra cable.
Satellites are a different matter. Since the satellite is the entire system (as opposed to a piece of the whole), it makes more and more sense to optimize for cost as launch costs come down since you can boost your reliability with spares.
Today's commercial and military systems already have spares. Launch costs will never come down to $10/pound, but if they did you could design to much less stringent criteria, because each satellite could 1) be heavier, 2) have a shorter lifespan, and 3) be less reliable as a result of redundancy. Look, what makes these things expensive is they must be as reliable as possible within the limits of our technology without weighing very much. You're paying ten times the cost for that extra digit.
It wouldn't take weeks or months, as I said, since you would have another one ready to power up. And while it's true you wouldn't build them on an assembly line, your design costs don't change if you build one, ten, or twenty. The second one is much cheaper than the first. In any event the primary reason we don't "need" many is because they cost too much. Lower the price point and you get more customers as the service enters their price range.
As far as space in GEO, that's nonsense. What's in short supply is space for active satellites, since they can't be so close they are indistinguishable to ground-based receivers. But there's plenty of space for "dark" birds. In any case you can use directional antennae to "multiplex" the slot, and for other comm applications lower orbits make sense anyway - with cheap phased array antennae GEO isn't as necessary as it once was, and it has its (latency) drawbacks.
~
They couldn't even make supersonic passenger jets economical. You can forget about scramjets.
When launch costs are lower you don't need satellites that last 15 years and tested to the nines. If I use the shuttle to launch it costs half a billion dollars to get my bird up (although a lot of that is picked up by the taxpayers), so I need something that's guarenteed to work. That means lots of expensive parts, and lots of expensive testing.
If I can launch cheaply I can afford to make cheaper satellites, since the cost of failure is lower. So now I need one less decimal place in my reliability, which means one less decimal place in the price. And I don't need the darned thing to work forever - a five year life might make more sense if I can replace it cheaply.
This makes the number of launches go up. Which makes the cost of the launch go down. Which makes the price of satellites go down. Take this loop a couple of times and you'll get closer to the actual production cost of the rocket, which is very low, in the grand scheme of things.
This research simply duplicates Austrailian research in 2001. Doesn't anybody remember the same furor at that time?