" So ARM has the know-how, the access and the pull to have a big say in what happens in the fabs roadmaps."
As the author:
I've talked to a number of people about this, on all sides of the equation: Foundry folks, Intel engineers, ARM, etc.
The universal consensus is that yes, Intel has an integration advantage. Even ARM admits that. Being able to directly tie silicon and process implementation to SoC architecture is a real benefit. On top of that, Intel has a technology advantage.
That said, there's a *reason* why ARM now works more closely with major foundry partners than it used to. There's a reason why Nvidia has called for tighter integration of SoC design teams and the foundries they work at. There's a reason why TSMC and GF have begun talking up tighter integration on their end, as well. The goal is to create a design environment that captures more of the Intel model's benefits.
Going the foundry way gets you out from under a mountain of development and build-out costs. Going the Intel way gets you additional flexibility and customization. I don't think we'll be able to call out which approach was better until we see Bay Trail facing off against some of the second-generation Cortex-A15 / Swift / Krait-class hardware, and that's not going to happen until 2014.
So let's look at that. Keep in mind, they can't talk about anything that happens.
1). Google announces a surprise move to, say, Europe. They can't say why. But since the NSA has direct taps on all web traffic flowing into the United States, they can't continue serving US customers if they want to secure their data.
2). Google announces a new product for US customers. It's Gmail Encrypted, with all the problems associated with it. Note that this doesn't excuse them from legal liabillity; the NSA would undoubtedly take them to court to argue that they were deliberately evading responsibility.
3). There's no way to just 'switch' incorporation. You have discorporate in the US and reincorporate in Europe. So you pay billions in capital gains on the US revenue, then hundreds of millions in associated fees and start-up costs. You can't just sell a company as big as Google to a European competitor and you can't just pick up the assets and move them, scot-free. In fact, the only way to really protect your data centers is to move them. All of them.
4). You can't tell anyone WHY you are doing this. You're just doing it.
How do you think Google's shareholders would have reacted to an announcement that Google was going to move to a European nation, dissolve its US corporation, move all its physical datacenters across the pond, institute a new mandatory encryption service for US customers, and pay enormous costs for doing so -- because it felt like a good idea?
Google is the company pressing in court to be able to talk about NSA gag letters. They were doing it, Pre-Snowden. That's not significant?
The bigger point, however, is that Google didn't have a choice. Microsoft didn't have a choice. Yahoo didn't get a choice. And if the NSA/FBI start gunning for Mozilla, Mozilla won't have a choice, either..
I've got no problem with your browser choice -- if you want to use Mozilla over Chrome, or IE over Firefox, hey, that's your call. But don't misrepresent the situation.
Google and Yahoo both pushed back hard against the NSA's programs. Yahoo went to court over it. You know what the court said? "Obey."
So what could Google do? You can't run an advertising business without having some information on your users. You can't run an email service without having access to the accounts. Yes, I suppose Google could have theoretically attempted to create a business in which everyone it served were direct customers of encryption services it provided (while explicitly saying that it couldn't decrypt traffic). Maybe that works for a startup, but you can't exactly transition a multi-billion dollar corporation to a direct customer model to avoid the NSA -- especially when you are legally prohibited from acknowledging that the NSA even spoke to you.
More than one of the companies that participate in Prism were forced to do so.
To get around really annoying types of ads that mandate that you sign up / sign in for a service. If I wanted to sign up for your website, I'd do it. Attempting to force me to do so for the sake of gathering better target data isn't interesting to me. Trying to force me to do so with giant ads, even less so.
There will be other ways to temporarily disable JS, so I'm not too worried about FF removing the chekc box. But it's annoying.
So people will have to learn early to treat their work/school accounts as non-private. That's a burden to put on school kids, I agree. And parents should be aware of it. But I don't think it's a five-alarm issue if discussed properly.
I seriously doubt the government mines the Internet accounts of seven year olds. Nevertheless, if it chooses to do so, not having school accounts won't stop it.
This is a known issue with both NV and AMD cards. Old games aren't tested to ensure compatibility with new drivers. It's not surprising that a nine year old game had trouble on a modern card in an ancient operating system. If you name the title, I might be able to dig up some advice on it.
"Send you child to a different school, speak with the school board, whatever it takes; you the parent have complete control..."
You're assuming a level of income and engagement that only exists at the high level. 1). The parents have to be educated on these issues *specifically*. 2). The parents need to have the money to make the changes you suggest.
You can't just change school districts. Generally, you're assigned a school based on where you live. Sending your kid to a different school means paying a penalty, at minimum, or paying for private school, where tuition can approach college-level. Can you afford to lay out $8,000 - $12,000 a year for your kid to go to a different school, while still paying property taxes to support the local ones? Keep in mind, you have to handle transportation to and from the school as well, which again, assumes you're rich enough to do so .
Of course there's home schooling -- provided, again, that you're rich enough to be able to afford not to work or have a spouse who can support your family on a single income. And it's a lot harder to be as engaged as a parent if you're the only one earning an income, leaving the house at 7 AM, and getting home 12 hours later to young kids who still need dinner made and homework checked.
All of *this* assumes that the school district itself properly understands the programs in question well enough to communicate them and that the programs are administered appropriately.
This question needs a bit more detail. What *are* the implications of using these Google services? Is Google using the same boilerplate contract? Does it sweep emails for words and phrases to show advertising? Is it collecting anonymous data?
I think you probably need some school-specific clauses to address the particular privacy and safeguards but you haven't articulated any specific examples of areas where you think Google is falling short or why this might become a problem. Kids are going to have digital footprints as children. I might not like that very much, and as a parent I may try to limit it, but you can't stop it.
Is a site with sexual advice "adult?" What about explicit sexual advice? What about discussion of non-normative sexuality (LBGT, BDSM, etc)? Does adult mean "Pornographic?" It's a ridiculously overbroad policy that's been horribly communicated. No one is arguing that Google doesn't have the right to make changes to its own services, but what the hell does or doesn't constitute "adult?"
On the one hand, good. Patients deserve to KNOW if their doctor fucked something up. Every now and then you hear horror stories about sponges, clamps, and god knows what else being left inside a patient, or a doctor that removes the wrong body part. Video playback could also help in a malpractice defense in which the patient claimed the doctor was distracted, intoxicated, or made a critical error.
On the other hand, knowing that there's a camera and live feed watching your every move isn't something I'd want to deal with while I was elbow deep in someone's gizzard.
The act of observing something changes the behavior of the people being observed. I'm not sold on this, save in particular training circumstances.
The question isn't "Can you get a bit-perfect result if you perfectly replicate everything" the question is: "Real-world people working with real-world compilers are likely to see results obfuscated as a result of different compiler versions, flags, and optimization options long before you get to the question of time stamps and offsets."
Unless I'm missing something pretty profound, even having the exact *source* won't always result in the exact binary. My understanding (and I could be wrong about this) is that you can take a well written program and plug it into multiple compilers. GCC may be one of the most popular options, but it's not the only one.
But compilers all optimize differently. GCC 3.x optimizes somewhat differently than GCC 4.x. You can tweak this behavior by manually setting compiler flags, or you can compile binaries that explicitly target different CPU architectures. A binary compiled to target all x86 processors may run differently on Haswell than a binary that's compiled specifically for Haswell.
In other words, flags set at compile time will change performance characteristics, even if the source code is identical, and while some projects may publish the exact details of every compiler flag they set, this doesn't seem to be the norm. Most projects I've seen say "Here are some binaries, and here's the source code if you want to play with it."
Clearly, the point of source code isn't to exactly duplicate every binary in every situation but to give you the data that goes *into* the compiler before the executable is compiled.
I assume the Tor story is legit, since it came out publicly that they traced him when he stayed in such-and-such a hotel. If he hadn't stayed there, he could've simply said so.
Plus, given the penchant for national security these days, the FBI didn't *have* to say anything. So the fact that they were willing to give a reason as opposed to silence, when silence has become so accepted, suggests they told the truth.
The problem with heavily encrypted solutions is that they rely on human perfection. There was a story a few months back about Sabu. He eluded the FBI for months until, in a hotel room, he made the mistake of logging into IRC without using Tor first.
That was all it took. One non-Tor login, and the FBI had him.
Human beings are not designed for constant watchfulness. We make mistakes. We screw up. Even if *you* stay perfect, the person or persons you're communicating with may not, and if the FBI or NSA wants the details of what you're talking about, they can "break" the encryption at either end of the conversation. Maybe they can't find you -- but if they find the people you're talking to, they can still grab the info.
I'm not saying that all security is useless, or that there's no benefit to raising the bar. My point is that the solution to this is to *stop spying.* Because, in the long run, almost everyone screws up.
What's your distinction between NAS and SAN? When I think SAN, I think Fibre Channel and server-side deployments. That's why I blinked when you talked about buying a SAN at OfficeMax. Buffalo doesn't seem to sell any SAN hardware, but NAS hardware is plentiful.
If you're calling SAN what I'm calling NAS, the price difference explains itself. As for boot times and the SSD/HDD difference, you've said you use SSDs for caching, which gets you much of the speed benefit.
As for gigabit "always" being enough, I don't know that I'd say *that*, always being a very long time. But with consumer demand pushing hard at *wireless* technology, and wireless being inevitably slower than wired, I think it's fair to say that current trends in consumer technology are hitting bottlenecks in wireless transfer speeds, battery power, and in some cases, ease-of-use. Windows 8 Storage Spaces is an example of a technology I wish was much easier to configure and maintain than it is.
Here's what I'd ask you, since you've got more experience with this type of network setup than I do: The max limit on GigE is 125MB/s. Overhead being what it is, I'd be surprised if two single-drive, conventional HDD systems can sustain that, even if one of them uses a boot drive to keep everything responsive (the data transfer itself is going from 1x HDD to 1x HDD).
Obviously a RAID 0 will boost mechanical transfer speeds, but at what point does 10gigE start yielding practical benefits for even fairly technical users? 1,250MB/s of bandwidth is great in theory, but that's faster than any four hard drives -- even hard drives in RAID 0 -- can write.
It seems to me that the value of the bandwidth diminishes as the bandwidth outpaces the maximum speed of data movement between two physical spinning disks on the one hand and the Internet pipeline on the other. If none of the content you're moving can sustain anything like 10gigE maximum bandwidth, what's the value of the unused capacity?
"The only reason they're not all SSD's is because SSD's are expensive and unreliable."
This was true years ago. I'm surprised you still think so now. I'm not saying SSDs make sense for every use case, but 300MB+ sustained read/write speed is within single-drive capabilities now, with drive costs down to $1 per GB for good MLC.
"normal people have been buying inexpensive Buffalo and WD SAN solutions at the office supply store since 2008."
I think that depends on your definition of normal and inexpensive. I'm somewhat surprised to see you declaring SSDs expensive, but SAN (even low-end SAN) as affordable. But we clearly have different use-cases. I have one system and a NAS device. 10GigE is only useful if the Internet pipeline supports it; my NAS is a two-drive solution that couldn't handle anything about gigabit, even if I had a hankering to play with it.
(I configure and test software/hardware on a regular basis, but these systems are always temporary with temporary OS installations.
Gaming. Even if Linux was truly an equal option, I don't *want* to go through the hassle of trying to get games working in Linux from years ago.
Gaming, however, is also what's keeping me on Windows *7.* I've found far too many Steam titles are a bit finicky with Windows 8 -- not so much that they don't work, period, as that getting them working turned into a major hassle, with GfW patches required and a great deal of hoop-jumping. I had trouble with both DiRT 3 and Arkham City on W8, while both were flawless in W7.
Beyond gaming, Linux, at best, is a "me-too" solution. Yes, Linux can be configured to do everything I need from Windows outside the gaming question. But it takes me far more time to learn to do that than it does to just use Windows.
That's always been the problem, from my perspective. Not that Linux is bad, or incompatible, but that it's never offered me enough of a personal advantage to make it worth switching. Obviously there are millions of people who have felt very differently, including plenty at Slashdot, and I've got *no* bone to pick with that. For whatever it's worth, I was unimpressed with Mac OS X 10.5 when I switched to it on a work laptop. After years of hype, I expected to be wow'd by this fundamentally amazing way of doing things. Instead, I discovered that Macs were organized on different principles and that I liked some of it and didn't like some of the rest.
Microsoft would have to fundamentally *break* something in Windows for me to switch. Alternately, Linux would have to add something amazing and unique that I couldn't get from an MS operating system. "Just as good" is never good enough when the cost of switching is a significant time and energy investment.
I'm not a cheapskate. I see no reason to invest in hardware that doesn't provide a tangible benefit to me.
That's not the same as saying "No one needs 10gigE." It means *I* don't. If I could get FiOS or Google Fiber, I'd be a lot more interested. But I'm already stuck at about 20Mbps and none of my systems have storage systems that can write at 1,250MB/s that 10GigE makes possible. Adopting it requires new routers and new Ethernet cards.
I'm not saying it's a bad standard. I'm not saying some people can't use it. But for the regular guy consumer, it's not very useful.
" Boot every PC from a remote server. No need even for local swap"
10GBe won't be as fast as a nice cheap SSD -- but not even an SSD can keep up with an avalanche of data requests from multiple systems unless that remote server is pretty damn beefy by home standards. Managing simultaneous uploads and downloads from multiple systems to a single home location won't be much fun. Simpler to keep your OS local, and trivial as far as cost.
* Have a much better time doing backups. True, provided your storage can handle it. Maximum read/write on 10gigE is 1,250MB/s. You'll need a heck of a RAID array for that, but it's buildable. Or, you could just stick with GigE, since that still tops out at 125MB/s and that pushes local (non-SSD) storage.
* Never worry about traffic spikes increasing latency on sensitive protocols. Correct me if I'm wrong, but my understanding is that this requires a lot more than just a high-speed connection. High-end connection + craptastic router = terrible latency when dealing with high load.
What you're talking about actually sounds like this:
"I have my home wired up like a datacenter. Everyone else should want a huge amount of network capacity and capability so that it makes my already extravagant costs slightly cheaper."
I'm totally in favor of that building code rewrite, so long as you're paying the enormous demolition bill. Costs vary widely depending on location, but the Internet projects a cost of $4000 - $7000 for a small house, $9,000 - $22,000 if you have to tear the foundation out.
Somehow I doubt that families still getting back on their feet after the 2008 recession have the funds to pay for an arbitrary knockdown and rebuild. Even when it might make sense to do so. Even given the damage that the tornadoes have done. If you don't have much money (and Oklahoma isn't exactly known for its billionaires), you go elsewhere before you tear down and start over.
Slashdot Mirror
" So ARM has the know-how, the access and the pull to have a big say in what happens in the fabs roadmaps."
As the author:
I've talked to a number of people about this, on all sides of the equation: Foundry folks, Intel engineers, ARM, etc.
The universal consensus is that yes, Intel has an integration advantage. Even ARM admits that. Being able to directly tie silicon and process implementation to SoC architecture is a real benefit. On top of that, Intel has a technology advantage.
That said, there's a *reason* why ARM now works more closely with major foundry partners than it used to. There's a reason why Nvidia has called for tighter integration of SoC design teams and the foundries they work at. There's a reason why TSMC and GF have begun talking up tighter integration on their end, as well. The goal is to create a design environment that captures more of the Intel model's benefits.
Going the foundry way gets you out from under a mountain of development and build-out costs. Going the Intel way gets you additional flexibility and customization. I don't think we'll be able to call out which approach was better until we see Bay Trail facing off against some of the second-generation Cortex-A15 / Swift / Krait-class hardware, and that's not going to happen until 2014.
As the author of the story: It's possible Anand and I were at the same meeting. ;)
So let's look at that. Keep in mind, they can't talk about anything that happens.
1). Google announces a surprise move to, say, Europe. They can't say why. But since the NSA has direct taps on all web traffic flowing into the United States, they can't continue serving US customers if they want to secure their data.
2). Google announces a new product for US customers. It's Gmail Encrypted, with all the problems associated with it. Note that this doesn't excuse them from legal liabillity; the NSA would undoubtedly take them to court to argue that they were deliberately evading responsibility.
3). There's no way to just 'switch' incorporation. You have discorporate in the US and reincorporate in Europe. So you pay billions in capital gains on the US revenue, then hundreds of millions in associated fees and start-up costs. You can't just sell a company as big as Google to a European competitor and you can't just pick up the assets and move them, scot-free. In fact, the only way to really protect your data centers is to move them. All of them.
4). You can't tell anyone WHY you are doing this. You're just doing it.
How do you think Google's shareholders would have reacted to an announcement that Google was going to move to a European nation, dissolve its US corporation, move all its physical datacenters across the pond, institute a new mandatory encryption service for US customers, and pay enormous costs for doing so -- because it felt like a good idea?
Google is the company pressing in court to be able to talk about NSA gag letters. They were doing it, Pre-Snowden. That's not significant?
The bigger point, however, is that Google didn't have a choice. Microsoft didn't have a choice. Yahoo didn't get a choice. And if the NSA/FBI start gunning for Mozilla, Mozilla won't have a choice, either..
I've got no problem with your browser choice -- if you want to use Mozilla over Chrome, or IE over Firefox, hey, that's your call. But don't misrepresent the situation.
Google and Yahoo both pushed back hard against the NSA's programs. Yahoo went to court over it. You know what the court said? "Obey."
So what could Google do? You can't run an advertising business without having some information on your users. You can't run an email service without having access to the accounts. Yes, I suppose Google could have theoretically attempted to create a business in which everyone it served were direct customers of encryption services it provided (while explicitly saying that it couldn't decrypt traffic). Maybe that works for a startup, but you can't exactly transition a multi-billion dollar corporation to a direct customer model to avoid the NSA -- especially when you are legally prohibited from acknowledging that the NSA even spoke to you.
More than one of the companies that participate in Prism were forced to do so.
To get around really annoying types of ads that mandate that you sign up / sign in for a service. If I wanted to sign up for your website, I'd do it. Attempting to force me to do so for the sake of gathering better target data isn't interesting to me. Trying to force me to do so with giant ads, even less so.
There will be other ways to temporarily disable JS, so I'm not too worried about FF removing the chekc box. But it's annoying.
So people will have to learn early to treat their work/school accounts as non-private. That's a burden to put on school kids, I agree. And parents should be aware of it. But I don't think it's a five-alarm issue if discussed properly.
I seriously doubt the government mines the Internet accounts of seven year olds. Nevertheless, if it chooses to do so, not having school accounts won't stop it.
This is a known issue with both NV and AMD cards. Old games aren't tested to ensure compatibility with new drivers. It's not surprising that a nine year old game had trouble on a modern card in an ancient operating system. If you name the title, I might be able to dig up some advice on it.
"Send you child to a different school, speak with the school board, whatever it takes; you the parent have complete control..."
You're assuming a level of income and engagement that only exists at the high level. 1). The parents have to be educated on these issues *specifically*. 2). The parents need to have the money to make the changes you suggest.
You can't just change school districts. Generally, you're assigned a school based on where you live. Sending your kid to a different school means paying a penalty, at minimum, or paying for private school, where tuition can approach college-level. Can you afford to lay out $8,000 - $12,000 a year for your kid to go to a different school, while still paying property taxes to support the local ones? Keep in mind, you have to handle transportation to and from the school as well, which again, assumes you're rich enough to do so .
Of course there's home schooling -- provided, again, that you're rich enough to be able to afford not to work or have a spouse who can support your family on a single income. And it's a lot harder to be as engaged as a parent if you're the only one earning an income, leaving the house at 7 AM, and getting home 12 hours later to young kids who still need dinner made and homework checked.
All of *this* assumes that the school district itself properly understands the programs in question well enough to communicate them and that the programs are administered appropriately.
This question needs a bit more detail. What *are* the implications of using these Google services? Is Google using the same boilerplate contract? Does it sweep emails for words and phrases to show advertising? Is it collecting anonymous data?
I think you probably need some school-specific clauses to address the particular privacy and safeguards but you haven't articulated any specific examples of areas where you think Google is falling short or why this might become a problem. Kids are going to have digital footprints as children. I might not like that very much, and as a parent I may try to limit it, but you can't stop it.
Precisely. What's "adult?"
Is a site with sexual advice "adult?" What about explicit sexual advice? What about discussion of non-normative sexuality (LBGT, BDSM, etc)? Does adult mean "Pornographic?" It's a ridiculously overbroad policy that's been horribly communicated. No one is arguing that Google doesn't have the right to make changes to its own services, but what the hell does or doesn't constitute "adult?"
We're at 5.4% of exaflop scale. Somehow I don't think this is a 2013 / 2014 goal ;)
The fact that video can be used as proof, in any direction (innocence or guilty) is not a reason not to video, ever.
I'm torn on that one.
On the one hand, good. Patients deserve to KNOW if their doctor fucked something up. Every now and then you hear horror stories about sponges, clamps, and god knows what else being left inside a patient, or a doctor that removes the wrong body part. Video playback could also help in a malpractice defense in which the patient claimed the doctor was distracted, intoxicated, or made a critical error.
On the other hand, knowing that there's a camera and live feed watching your every move isn't something I'd want to deal with while I was elbow deep in someone's gizzard.
The act of observing something changes the behavior of the people being observed. I'm not sold on this, save in particular training circumstances.
Sure. But that's not my point.
The question isn't "Can you get a bit-perfect result if you perfectly replicate everything" the question is: "Real-world people working with real-world compilers are likely to see results obfuscated as a result of different compiler versions, flags, and optimization options long before you get to the question of time stamps and offsets."
Nothing you said indicates this *isn't* true.
Unless I'm missing something pretty profound, even having the exact *source* won't always result in the exact binary. My understanding (and I could be wrong about this) is that you can take a well written program and plug it into multiple compilers. GCC may be one of the most popular options, but it's not the only one.
But compilers all optimize differently. GCC 3.x optimizes somewhat differently than GCC 4.x. You can tweak this behavior by manually setting compiler flags, or you can compile binaries that explicitly target different CPU architectures. A binary compiled to target all x86 processors may run differently on Haswell than a binary that's compiled specifically for Haswell.
In other words, flags set at compile time will change performance characteristics, even if the source code is identical, and while some projects may publish the exact details of every compiler flag they set, this doesn't seem to be the norm. Most projects I've seen say "Here are some binaries, and here's the source code if you want to play with it."
Clearly, the point of source code isn't to exactly duplicate every binary in every situation but to give you the data that goes *into* the compiler before the executable is compiled.
Or am I missing something?
I assume the Tor story is legit, since it came out publicly that they traced him when he stayed in such-and-such a hotel. If he hadn't stayed there, he could've simply said so.
Plus, given the penchant for national security these days, the FBI didn't *have* to say anything. So the fact that they were willing to give a reason as opposed to silence, when silence has become so accepted, suggests they told the truth.
The problem with heavily encrypted solutions is that they rely on human perfection. There was a story a few months back about Sabu. He eluded the FBI for months until, in a hotel room, he made the mistake of logging into IRC without using Tor first.
That was all it took. One non-Tor login, and the FBI had him.
Human beings are not designed for constant watchfulness. We make mistakes. We screw up. Even if *you* stay perfect, the person or persons you're communicating with may not, and if the FBI or NSA wants the details of what you're talking about, they can "break" the encryption at either end of the conversation. Maybe they can't find you -- but if they find the people you're talking to, they can still grab the info.
I'm not saying that all security is useless, or that there's no benefit to raising the bar. My point is that the solution to this is to *stop spying.* Because, in the long run, almost everyone screws up.
What's your distinction between NAS and SAN? When I think SAN, I think Fibre Channel and server-side deployments. That's why I blinked when you talked about buying a SAN at OfficeMax. Buffalo doesn't seem to sell any SAN hardware, but NAS hardware is plentiful.
If you're calling SAN what I'm calling NAS, the price difference explains itself. As for boot times and the SSD/HDD difference, you've said you use SSDs for caching, which gets you much of the speed benefit.
As for gigabit "always" being enough, I don't know that I'd say *that*, always being a very long time. But with consumer demand pushing hard at *wireless* technology, and wireless being inevitably slower than wired, I think it's fair to say that current trends in consumer technology are hitting bottlenecks in wireless transfer speeds, battery power, and in some cases, ease-of-use. Windows 8 Storage Spaces is an example of a technology I wish was much easier to configure and maintain than it is.
Here's what I'd ask you, since you've got more experience with this type of network setup than I do: The max limit on GigE is 125MB/s. Overhead being what it is, I'd be surprised if two single-drive, conventional HDD systems can sustain that, even if one of them uses a boot drive to keep everything responsive (the data transfer itself is going from 1x HDD to 1x HDD).
Obviously a RAID 0 will boost mechanical transfer speeds, but at what point does 10gigE start yielding practical benefits for even fairly technical users? 1,250MB/s of bandwidth is great in theory, but that's faster than any four hard drives -- even hard drives in RAID 0 -- can write.
It seems to me that the value of the bandwidth diminishes as the bandwidth outpaces the maximum speed of data movement between two physical spinning disks on the one hand and the Internet pipeline on the other. If none of the content you're moving can sustain anything like 10gigE maximum bandwidth, what's the value of the unused capacity?
"The only reason they're not all SSD's is because SSD's are expensive and unreliable."
This was true years ago. I'm surprised you still think so now. I'm not saying SSDs make sense for every use case, but 300MB+ sustained read/write speed is within single-drive capabilities now, with drive costs down to $1 per GB for good MLC.
"normal people have been buying inexpensive Buffalo and WD SAN solutions at the office supply store since 2008."
I think that depends on your definition of normal and inexpensive. I'm somewhat surprised to see you declaring SSDs expensive, but SAN (even low-end SAN) as affordable. But we clearly have different use-cases. I have one system and a NAS device. 10GigE is only useful if the Internet pipeline supports it; my NAS is a two-drive solution that couldn't handle anything about gigabit, even if I had a hankering to play with it.
(I configure and test software/hardware on a regular basis, but these systems are always temporary with temporary OS installations.
Gaming. Even if Linux was truly an equal option, I don't *want* to go through the hassle of trying to get games working in Linux from years ago.
Gaming, however, is also what's keeping me on Windows *7.* I've found far too many Steam titles are a bit finicky with Windows 8 -- not so much that they don't work, period, as that getting them working turned into a major hassle, with GfW patches required and a great deal of hoop-jumping. I had trouble with both DiRT 3 and Arkham City on W8, while both were flawless in W7.
Beyond gaming, Linux, at best, is a "me-too" solution. Yes, Linux can be configured to do everything I need from Windows outside the gaming question. But it takes me far more time to learn to do that than it does to just use Windows.
That's always been the problem, from my perspective. Not that Linux is bad, or incompatible, but that it's never offered me enough of a personal advantage to make it worth switching. Obviously there are millions of people who have felt very differently, including plenty at Slashdot, and I've got *no* bone to pick with that. For whatever it's worth, I was unimpressed with Mac OS X 10.5 when I switched to it on a work laptop. After years of hype, I expected to be wow'd by this fundamentally amazing way of doing things. Instead, I discovered that Macs were organized on different principles and that I liked some of it and didn't like some of the rest.
Microsoft would have to fundamentally *break* something in Windows for me to switch. Alternately, Linux would have to add something amazing and unique that I couldn't get from an MS operating system. "Just as good" is never good enough when the cost of switching is a significant time and energy investment.
I'm not a cheapskate. I see no reason to invest in hardware that doesn't provide a tangible benefit to me.
That's not the same as saying "No one needs 10gigE." It means *I* don't. If I could get FiOS or Google Fiber, I'd be a lot more interested. But I'm already stuck at about 20Mbps and none of my systems have storage systems that can write at 1,250MB/s that 10GigE makes possible. Adopting it requires new routers and new Ethernet cards.
I'm not saying it's a bad standard. I'm not saying some people can't use it. But for the regular guy consumer, it's not very useful.
" Boot every PC from a remote server. No need even for local swap"
10GBe won't be as fast as a nice cheap SSD -- but not even an SSD can keep up with an avalanche of data requests from multiple systems unless that remote server is pretty damn beefy by home standards. Managing simultaneous uploads and downloads from multiple systems to a single home location won't be much fun. Simpler to keep your OS local, and trivial as far as cost.
* Have a much better time doing backups.
True, provided your storage can handle it. Maximum read/write on 10gigE is 1,250MB/s. You'll need a heck of a RAID array for that, but it's buildable. Or, you could just stick with GigE, since that still tops out at 125MB/s and that pushes local (non-SSD) storage.
* Never worry about traffic spikes increasing latency on sensitive protocols.
Correct me if I'm wrong, but my understanding is that this requires a lot more than just a high-speed connection. High-end connection + craptastic router = terrible latency when dealing with high load.
What you're talking about actually sounds like this:
"I have my home wired up like a datacenter. Everyone else should want a huge amount of network capacity and capability so that it makes my already extravagant costs slightly cheaper."
I'm totally in favor of that building code rewrite, so long as you're paying the enormous demolition bill. Costs vary widely depending on location, but the Internet projects a cost of $4000 - $7000 for a small house, $9,000 - $22,000 if you have to tear the foundation out.
Somehow I doubt that families still getting back on their feet after the 2008 recession have the funds to pay for an arbitrary knockdown and rebuild. Even when it might make sense to do so. Even given the damage that the tornadoes have done. If you don't have much money (and Oklahoma isn't exactly known for its billionaires), you go elsewhere before you tear down and start over.