I think these PCIe flash drives win for raw performance because they have access to the entire device at PCIe bus speeds.
Maybe some ideal 16x card with a dedicated SATA controller per connection would give you RAID-0 performance of 30 Gbits/sec for five disks, but something tells me you'd be limited by the individual SATA limit of 6 Gbit/sec. In the real world, I don't know that any $100 8x RAID card would do that.
But I would also bet that most workloads are IOP bound, not mass throughput bound, and with reasonable amounts of RAID cache on the card there would be little practical difference between the two.
Is it really the money or is it the background check nonsense that scares people away?
I would think the latter would be a big influence. Even if you had no serious skeletons in your closet (no arrests, not a drug user, etc) there's still a certain paranoia that the FBI is asking a lot of people a lot of questions. And who knows what some asshole that doesn't like you might say?
And MOST people have some kind of skeleton in their closet (smoke/smoked pot, some kind of sex thing, whatever).
It'd be curious to compare Goldman Sachs policies with the NSAs for technology. Both want the best and brightest, both deal with sensitive areas (sure, maybe National Security is higher but so is program trading with billions of dollars), but is Goldman going to turn down some eccentric with a PhD in math with a deep knowledge of modeling because he smokes pot?
If an ocean liner is the Waldorf-Astoria, a think an airship would end up being more Holiday Inn express. From what I've read, the Hindenburg was pretty spartan in terms of accommodations, especially in comparison to the liners of its era.
The Hindenburg was a lot bigger than the Airlander and carried a maximum of 72 passengers. It's hard to see the smaller Airlander carrying more than 30 passengers, maybe less depending on the level of amenities and size of berths. Carrying 2-300 passengers would seem like it would take a massive airship that would make the 800 ft long Hindenburg seem tiny.
Natalie Portman was great in "Closer" and "Black Swan", of the those you group her with, she's probably the best actress.
I don't think Scarlett is all that good, her initial success in "Lost in Translation" seems like a fluke. But she's mostly managed to turn herself into an action babe, so I'd guess she's realized that drama isn't her thing.
Mila Kunis, good or bad, is something of a curiosity. She was a sitcom bimbo but has had a turn of fairly decent acting with Portman in "Black Swan" but then took a turn for more mundane stuff.
Megan Fox is just a pretty face. Jennifer Lawrence is pretty good, but her naive response to the leaking of her nude photos was tedious.
Meryl Streep is good, but after a while she kind of plays Meryl Streep or at least its hard to not see her as Meryl Streep Playing Her Character.
Helen Mirren is great, but was she always great or did she become great after "Elizabeth I" late in her career? It's hard to think of anything memorable in her career prior to "Cook, the Thief.." and her turn on the cop drama "Prime Suspect".
...then you're just going to be buttfucked by the ones who get up to mischief before they resign. You should have the ability built-in to recover from whatever they do, whenever they do it, because the worst damage is done by the insider you never suspect.
I'm generally in favor of the idea of that once someone submits a resignation, you might as well just tell them they don't need to come in. They can't get anything meaningful done in two weeks anyway and if you "need" them to explain what they do/project status/etc, then you're doing it wrong anyway and you won't find two weeks nearly enough time to get caught up.
Plus, what kind of leverage do you hold over someone who quit and has a job, anyway? Short of criminal behavior, you've got none. I've known a couple of managers at companies I worked at who were total assholes to employees who left, demanding extra work, tons of documentation, etc. It baffled me why the employees put up with it and knowing one manager in particular, I'm sure her employees hated her anyway and fucked up the work she made them do anyway. I know I heard rumors of shredded original billing materials and other documentation.
If you're desperate for a resignees information and talents, the best choice is to offer them a consultancy contract for real money. I think this gets people's respect, real quick. It shows you actually value their knowledge and skills (versus some bullshit words) and it buys you some leverage, since no work == no pay. But it has to be real money and guaranteed, "we might want you back for something later..." is no more believable than "let's have sex tomorrow instead." Tomorrow never comes.
The notion that there is some kind of Gentleman's Rules surrounding employment is over. Everyone knows they can be axed at the drop of a hat and most people feel no loyalty to their employer (or shouldn't, anyway) and could walk tomorrow. You have to be prepared now, not when they leave.
but why should a minority of us suffer due to a majority that aren't capable to make their own choices?
How is that not true of pretty much anything that has risk/danger associated with it which is ameliorated by prudence and caution?
Drugs: Many people are capable of using drugs sanely without risking themselves or other people, but because some minority shows absolutely no control we have massive controls on drugs.
Weapons: Many people are perfectly capable of safely owning even very destructive weapons without hurting themselves or others. But because some minority of people do batshit crazy things with weapons, we have a lot of controls on gun ownership and extreme controls on certain types of guns (automatic weapons, etc).
The list is endless. A minority of people are stupid, lack self control and any kind of prudence so we implement controls which address the lowest common denominator, occasionally allowing some people to jump through hoops to obtain slightly more access to something, but often with another set of draconian controls applied.
Forget these two guys and their bitcoin score, how much CASH walks away during drug investigations? How much is outright stolen, how much is extorted? How much is taken in product in lieu of cash?
This is one of the most pernicious aspects to drug criminalization, the huge potential for corruption by law enforcement.
And it's just another problem completely eliminated by legalization.
The language always seems kind of inflammatory, but sometimes I think they have something of a point.
When calculating risks and outcomes, everybody brings certain biases to the table about what are considered acceptable outcomes, losses and gains. That those biases may be driven by "masculinized rationality" may be taking it a bit far, but the idea that it's not a perfectly bright line threshold and that some tradeoffs may be involved shouldn't be disregarded.
The irony is that the same logic applied to the job by the worker basically means -- I'm free to do whatever I want at this job, and if it doesn't work out of them they can fire me.
For the company, the logic means they can be abusive, discriminatory, dishonest and exploitive.
So for the worker then, I guess they can be lazy, dishonest, unproductive, etc. It's the worker's role to exploit the company for the maximum gain they can get. Maximum shirk, minimum work.
What's funny is, I would bet that author if presented with her own logic from a worker perspective would probably immediately launch into a diatribe about the worker's moral obligation to work hard, be a good employee, etc, yet she refuses to see any moral obligation by the employer to the employee.
My guess is that it's only a serious issue for people with specific IP knowledge, like higher-end people in pharma, chemicals, semiconductors, some kinds of software -- the kinds of skills with very limited places to use them, most with direct competitors.
For other jobs, like mostly generic IT work, I just can't see my boss bothering to spend the money to figure out where I might have moved to, provided I keep a low-ish profile about it.
Is it really "paranoia" (a mental disease involving ungrounded fears) if the fear is substantiated?
Are you now, or have you ever been, a member of the Communist party?
I'd say the number of non-threats who were actively and vigorously blackballed might call into question as to where the boundary between legitimate fear and paranoia fear is on this topic.
But, somehow, that clear and present danger of Communism no longer played the role it played during Korea War. Why?
Probably no one single answer. I don't think the early years of Viet Nam faced that much ideological opposition. I do think that the political-based mismanagement of the war led to "conventional" opposition to it. Then add in civil rights discontent, the exemptions that made it a "poor man's war" and the general social upheaval of the 1960s, shake well and pour over ice.
There was a lot of paranoia about Communist conspiracies. The Rosenberg trials. Joe McCarthy was making headlines "exposing" Communists. In some sense, there was some legitimate fear of Communist actions -- the Soviets had blockaded West Berlin, leading to the Berlin airlift in 1948.
Not only was the political climate dangerous for anyone opposing fighting Communist expansion in Korea, it wasn't irrational to believe that expansionist communism was a real threat, especially after recently fighting a war against two nations who started wars of imperial expansion, at least one of whom did so under the guise of a totalitarian political philosophy.
Sure, and I could also hotplug USB3 disks and cut even more power/space/complexity if I wanted to futz with turning it on and off.
Power cycling a NAS may be worthwhile if it's some kind of archive you don't use often but it doesn't make a ton of sense if you want it online more than offline.
The electricity cost is what's so painful. I have an old Intel Q6600 system with 4x1TB disks. Power consumption is something like 110 watts, or $11/month.
I could swap the drives out and keep power about the same, but at some point it becomes kind of expensive to keep spinning disks.
I think major leaps of density will eliminate platters. Why bother with them at all with their ridiculously slow seek times, heat, power consumption? At high capacities they're more of a risk to data integrity due to slow array rebuild times and it takes dozens of them to equal the IOPS of flash. Even now platters are either useful for their high density as Tier 3 in a SAN or in large numbers to get IOPS.
If there was a huge leap in flash densities I think they would get cheap enough that no one would bother, even if they were "unreliable" consumer MLC technology. Vendors could just double the extra flash used for recovery of bad cells and increase the endurance.
I think I've read that the Israelis have communicated back-channel to key actors that they will respond to a nuclear or chemical attack against Israeli with a response that will hit *all* major Arab capitals and Mecca.
To your larger point, I think only desperate, religiously motivated non-state actors reasonably believe that they can "get away" with use of a nuclear weapon. Either via subterfuge or because they believe in some kind of metaphysical redemption that transcends any material consequences.
I think even the worst bad state actors understand that state use of a nuclear weapon has a significant possibility of devastating retaliation which would end their state as they know it and possibly lead to the disintegration of the civilization it represents.
Think of the domestic political situation in the United States relative to being attacked with a nuclear weapon. For one, I would imagine that there would be significant demands within the military for a retaliatory nuclear strike as a preemption against a further strike. The American public would DEMAND a retaliatory strike and political pressure would very likely lead to one on its own.
In what fucking world do you think it would have ever been politically acceptable to allow the Japanese a negotiated surrender after 4 years of war and after Pearl Harbor Especially when that would have been approved by an unelected President like Truman?
I would imagine that the converse was true, that there were elements who wanted to *continue* nuking Japan after the second strike as retaliation for starting the war.
Heat is heat, it's maybe less efficient to redistribute it throughout a house than in a single room, but a rack of servers puts out a lot of heat.
You would want a thermostat that controls an input damper and an output damper, so that when it called for heat the servers recirculated the indoor air and when it didn't, the severs drew air from outside and output it outside. An existing furnace could provide supplementary heat if the rack's heat output wasn't sufficient.
I think the bigger idea has a lot of drawbacks.
Data connectivity? Maybe in the Netherlands everyone has access to gig fiber at residential addresses, but that wouldn't work in the US.
Regular server maintenance? Parts like disk drives break often enough that I wouldn't want to have to deal with the technician all the time, especially not off hours.
Power? At a residential address you would need some significant wiring done and a separate meter for the server rack. What about power outages?
But it does provide an esoteric data center model. With the right site selection, you could have a very distributed compute facility that would be insulated from single-site failures. But it would only work for the kinds of distributed workloads that don't care about a bunch of nodes dropping off.
Let me ask you this: if a country would come into the US and start razing cities and towns, would that break your will to fight? Or would that just inflame your desire to see of the invaders dead?
Of course, initially everyone has a natural response -- rally 'round the flag. Kill the invaders.
Now, what happens when people hear about the invasions continued advance? Cities in ruins, millions killed? Resisting military units wiped out, irregular paramilitary units crushed, cities and towns harboring resistance razed, their inhabitants summarily executed. Oh, and your town has been bombed, food supplies are sketchy, no electricity, etc.
Eventually the idea of anything but total surrender becomes impossible.
Our appetite for foreign militarism is entirely the result of our politicians selling the idea that our enemy is the leadership and their military forces, but the populace is our friend. With our advanced military weapons, we can defeat the defined "enemy" and then the populace will embrace us as liberators.
What I don't know is where this idea originated. My only guess was that it grew out of the reconstruction era in postwar Germany where civilian resistance was minimal and largely theoretical understandings of the Soviet domestic political climate.
Both of these seem naive. The Allies let the Germans starve for a couple of years after the war and most felt this was a better alternative than their experience with the Soviets. Despite Stalin and his repression, the Russians took massive losses and fought for the Soviet state. Much of this was compelled, but at the same time the populace did it.
Yet somehow, it's become a cornerstone of US military policy that the civilian population is at worst neutral and most likely supports US goals, not to mention US belief systems and values. Which is ironic if you look at most of the American and British propaganda from WWII, which sold the idea that the enemy nations were subhuman races which deserved to be wiped off the map.
It's been ages since I've used ipfw as a WAN simulator, but my memory of it is normally around a fairly static kind of configuration of latencies and bandwidth.
Simulating a cellular link that might hop between LTE and 1x kinds of data might be tough to do without some kind of engine which dynamically reprograms dummynets for vastly different bandwidth/latency scenarios to better simulate a node moving between 1x and LTE speeds. When I built a WAN simulator, I did to actually simulate known WAN link performance parameters.such as bandwidth and latency. I didn't have to worry about my link switching from multilink T1s to 512K frame relay to 56k dynamically.
They could have also provided a ton of statistical profile data so that the simulations closely mirrored real-world throughput associated with various media, especially common variability patterns.
A nice GUI front-end would be useful too, with actual throughput measured.
USB would not be desirable for internal system use, too much overhead. It is well designed for the purpose it has but you wouldn't want it for everything.
But what is "too much overhead" when the transport link gets fast enough? If USB4 ends up with 20 GBit/sec, overhead for anything but SAN shelf backplanes really won't matter.
I actually think I *would* want it for everything. One connector for disks and other peripherals, usable internally and externally. The way they package SSDs now you wouldn't even need to bother with an enclosure.
I think the real problem USB specifically has is a marginal performance history with USB2 devices (high CPU usage, low throughput) and Microsoft's steadfast refusal to allow Windows installs to USB devices, even USB3 (which makes no sense, really, when I can benchmark an ordinary PNY USB3 stick @ 110 MByte/sec read and 60 MByte/sec write).
If Windows could boot off USB3, eSATA would be largely forgotten as faster but with clunky, limited cabling and even SATA as a connector internal standard might get relegated to "enthusiast" boards where some minor performance boost was seen as valuable. M.SATA adoption would end up only in places where extreme miniaturization matters or the same enthusiast crowds.
I know it sounds crazy, but it sure seems a lot less crazy with USB3.1 supposed to hit 10 Gbits.
Slashdot Mirror
I think these PCIe flash drives win for raw performance because they have access to the entire device at PCIe bus speeds.
Maybe some ideal 16x card with a dedicated SATA controller per connection would give you RAID-0 performance of 30 Gbits/sec for five disks, but something tells me you'd be limited by the individual SATA limit of 6 Gbit/sec. In the real world, I don't know that any $100 8x RAID card would do that.
But I would also bet that most workloads are IOP bound, not mass throughput bound, and with reasonable amounts of RAID cache on the card there would be little practical difference between the two.
Is it really the money or is it the background check nonsense that scares people away?
I would think the latter would be a big influence. Even if you had no serious skeletons in your closet (no arrests, not a drug user, etc) there's still a certain paranoia that the FBI is asking a lot of people a lot of questions. And who knows what some asshole that doesn't like you might say?
And MOST people have some kind of skeleton in their closet (smoke/smoked pot, some kind of sex thing, whatever).
It'd be curious to compare Goldman Sachs policies with the NSAs for technology. Both want the best and brightest, both deal with sensitive areas (sure, maybe National Security is higher but so is program trading with billions of dollars), but is Goldman going to turn down some eccentric with a PhD in math with a deep knowledge of modeling because he smokes pot?
If an ocean liner is the Waldorf-Astoria, a think an airship would end up being more Holiday Inn express. From what I've read, the Hindenburg was pretty spartan in terms of accommodations, especially in comparison to the liners of its era.
The Hindenburg was a lot bigger than the Airlander and carried a maximum of 72 passengers. It's hard to see the smaller Airlander carrying more than 30 passengers, maybe less depending on the level of amenities and size of berths. Carrying 2-300 passengers would seem like it would take a massive airship that would make the 800 ft long Hindenburg seem tiny.
Natalie Portman was great in "Closer" and "Black Swan", of the those you group her with, she's probably the best actress.
I don't think Scarlett is all that good, her initial success in "Lost in Translation" seems like a fluke. But she's mostly managed to turn herself into an action babe, so I'd guess she's realized that drama isn't her thing.
Mila Kunis, good or bad, is something of a curiosity. She was a sitcom bimbo but has had a turn of fairly decent acting with Portman in "Black Swan" but then took a turn for more mundane stuff.
Megan Fox is just a pretty face. Jennifer Lawrence is pretty good, but her naive response to the leaking of her nude photos was tedious.
Meryl Streep is good, but after a while she kind of plays Meryl Streep or at least its hard to not see her as Meryl Streep Playing Her Character.
Helen Mirren is great, but was she always great or did she become great after "Elizabeth I" late in her career? It's hard to think of anything memorable in her career prior to "Cook, the Thief.." and her turn on the cop drama "Prime Suspect".
...blames his opponents, the officer corps, or some expatriate mullah hiding in America for this?
Or pays his cronies billions to "fix" the grid?
...then you're just going to be buttfucked by the ones who get up to mischief before they resign. You should have the ability built-in to recover from whatever they do, whenever they do it, because the worst damage is done by the insider you never suspect.
I'm generally in favor of the idea of that once someone submits a resignation, you might as well just tell them they don't need to come in. They can't get anything meaningful done in two weeks anyway and if you "need" them to explain what they do/project status/etc, then you're doing it wrong anyway and you won't find two weeks nearly enough time to get caught up.
Plus, what kind of leverage do you hold over someone who quit and has a job, anyway? Short of criminal behavior, you've got none. I've known a couple of managers at companies I worked at who were total assholes to employees who left, demanding extra work, tons of documentation, etc. It baffled me why the employees put up with it and knowing one manager in particular, I'm sure her employees hated her anyway and fucked up the work she made them do anyway. I know I heard rumors of shredded original billing materials and other documentation.
If you're desperate for a resignees information and talents, the best choice is to offer them a consultancy contract for real money. I think this gets people's respect, real quick. It shows you actually value their knowledge and skills (versus some bullshit words) and it buys you some leverage, since no work == no pay. But it has to be real money and guaranteed, "we might want you back for something later..." is no more believable than "let's have sex tomorrow instead." Tomorrow never comes.
The notion that there is some kind of Gentleman's Rules surrounding employment is over. Everyone knows they can be axed at the drop of a hat and most people feel no loyalty to their employer (or shouldn't, anyway) and could walk tomorrow. You have to be prepared now, not when they leave.
but why should a minority of us suffer due to a majority that aren't capable to make their own choices?
How is that not true of pretty much anything that has risk/danger associated with it which is ameliorated by prudence and caution?
Drugs: Many people are capable of using drugs sanely without risking themselves or other people, but because some minority shows absolutely no control we have massive controls on drugs.
Weapons: Many people are perfectly capable of safely owning even very destructive weapons without hurting themselves or others. But because some minority of people do batshit crazy things with weapons, we have a lot of controls on gun ownership and extreme controls on certain types of guns (automatic weapons, etc).
The list is endless. A minority of people are stupid, lack self control and any kind of prudence so we implement controls which address the lowest common denominator, occasionally allowing some people to jump through hoops to obtain slightly more access to something, but often with another set of draconian controls applied.
Forget these two guys and their bitcoin score, how much CASH walks away during drug investigations? How much is outright stolen, how much is extorted? How much is taken in product in lieu of cash?
This is one of the most pernicious aspects to drug criminalization, the huge potential for corruption by law enforcement.
And it's just another problem completely eliminated by legalization.
The language always seems kind of inflammatory, but sometimes I think they have something of a point.
When calculating risks and outcomes, everybody brings certain biases to the table about what are considered acceptable outcomes, losses and gains. That those biases may be driven by "masculinized rationality" may be taking it a bit far, but the idea that it's not a perfectly bright line threshold and that some tradeoffs may be involved shouldn't be disregarded.
The irony is that the same logic applied to the job by the worker basically means -- I'm free to do whatever I want at this job, and if it doesn't work out of them they can fire me.
For the company, the logic means they can be abusive, discriminatory, dishonest and exploitive.
So for the worker then, I guess they can be lazy, dishonest, unproductive, etc. It's the worker's role to exploit the company for the maximum gain they can get. Maximum shirk, minimum work.
What's funny is, I would bet that author if presented with her own logic from a worker perspective would probably immediately launch into a diatribe about the worker's moral obligation to work hard, be a good employee, etc, yet she refuses to see any moral obligation by the employer to the employee.
"Any sufficiently advanced incompetence is indistinguishable from malice."
Sounds almost like a corollary to Clark's third law.
My guess is that it's only a serious issue for people with specific IP knowledge, like higher-end people in pharma, chemicals, semiconductors, some kinds of software -- the kinds of skills with very limited places to use them, most with direct competitors.
For other jobs, like mostly generic IT work, I just can't see my boss bothering to spend the money to figure out where I might have moved to, provided I keep a low-ish profile about it.
Is it really "paranoia" (a mental disease involving ungrounded fears) if the fear is substantiated?
Are you now, or have you ever been, a member of the Communist party?
I'd say the number of non-threats who were actively and vigorously blackballed might call into question as to where the boundary between legitimate fear and paranoia fear is on this topic.
But, somehow, that clear and present danger of Communism no longer played the role it played during Korea War. Why?
Probably no one single answer. I don't think the early years of Viet Nam faced that much ideological opposition. I do think that the political-based mismanagement of the war led to "conventional" opposition to it. Then add in civil rights discontent, the exemptions that made it a "poor man's war" and the general social upheaval of the 1960s, shake well and pour over ice.
It was vastly different political era.
There was a lot of paranoia about Communist conspiracies. The Rosenberg trials. Joe McCarthy was making headlines "exposing" Communists. In some sense, there was some legitimate fear of Communist actions -- the Soviets had blockaded West Berlin, leading to the Berlin airlift in 1948.
Not only was the political climate dangerous for anyone opposing fighting Communist expansion in Korea, it wasn't irrational to believe that expansionist communism was a real threat, especially after recently fighting a war against two nations who started wars of imperial expansion, at least one of whom did so under the guise of a totalitarian political philosophy.
Sure, and I could also hotplug USB3 disks and cut even more power/space/complexity if I wanted to futz with turning it on and off.
Power cycling a NAS may be worthwhile if it's some kind of archive you don't use often but it doesn't make a ton of sense if you want it online more than offline.
The electricity cost is what's so painful. I have an old Intel Q6600 system with 4x1TB disks. Power consumption is something like 110 watts, or $11/month.
I could swap the drives out and keep power about the same, but at some point it becomes kind of expensive to keep spinning disks.
I think major leaps of density will eliminate platters. Why bother with them at all with their ridiculously slow seek times, heat, power consumption? At high capacities they're more of a risk to data integrity due to slow array rebuild times and it takes dozens of them to equal the IOPS of flash. Even now platters are either useful for their high density as Tier 3 in a SAN or in large numbers to get IOPS.
If there was a huge leap in flash densities I think they would get cheap enough that no one would bother, even if they were "unreliable" consumer MLC technology. Vendors could just double the extra flash used for recovery of bad cells and increase the endurance.
I think I've read that the Israelis have communicated back-channel to key actors that they will respond to a nuclear or chemical attack against Israeli with a response that will hit *all* major Arab capitals and Mecca.
To your larger point, I think only desperate, religiously motivated non-state actors reasonably believe that they can "get away" with use of a nuclear weapon. Either via subterfuge or because they believe in some kind of metaphysical redemption that transcends any material consequences.
I think even the worst bad state actors understand that state use of a nuclear weapon has a significant possibility of devastating retaliation which would end their state as they know it and possibly lead to the disintegration of the civilization it represents.
Think of the domestic political situation in the United States relative to being attacked with a nuclear weapon. For one, I would imagine that there would be significant demands within the military for a retaliatory nuclear strike as a preemption against a further strike. The American public would DEMAND a retaliatory strike and political pressure would very likely lead to one on its own.
In what fucking world do you think it would have ever been politically acceptable to allow the Japanese a negotiated surrender after 4 years of war and after Pearl Harbor Especially when that would have been approved by an unelected President like Truman?
I would imagine that the converse was true, that there were elements who wanted to *continue* nuking Japan after the second strike as retaliation for starting the war.
"Parallels" could have been a Sliders type show if it had been picked up as a series and not released as an edited-into-a-movie pilot only on Netflix.
It was done by the guy who did "The Lost Room" which is probably one of the best things ever to appear on SciFi.
Heat is heat, it's maybe less efficient to redistribute it throughout a house than in a single room, but a rack of servers puts out a lot of heat.
You would want a thermostat that controls an input damper and an output damper, so that when it called for heat the servers recirculated the indoor air and when it didn't, the severs drew air from outside and output it outside. An existing furnace could provide supplementary heat if the rack's heat output wasn't sufficient.
I think the bigger idea has a lot of drawbacks.
Data connectivity? Maybe in the Netherlands everyone has access to gig fiber at residential addresses, but that wouldn't work in the US.
Regular server maintenance? Parts like disk drives break often enough that I wouldn't want to have to deal with the technician all the time, especially not off hours.
Power? At a residential address you would need some significant wiring done and a separate meter for the server rack. What about power outages?
But it does provide an esoteric data center model. With the right site selection, you could have a very distributed compute facility that would be insulated from single-site failures. But it would only work for the kinds of distributed workloads that don't care about a bunch of nodes dropping off.
Let me ask you this: if a country would come into the US and start razing cities and towns, would that break your will to fight? Or would that just inflame your desire to see of the invaders dead?
Of course, initially everyone has a natural response -- rally 'round the flag. Kill the invaders.
Now, what happens when people hear about the invasions continued advance? Cities in ruins, millions killed? Resisting military units wiped out, irregular paramilitary units crushed, cities and towns harboring resistance razed, their inhabitants summarily executed. Oh, and your town has been bombed, food supplies are sketchy, no electricity, etc.
Eventually the idea of anything but total surrender becomes impossible.
Our appetite for foreign militarism is entirely the result of our politicians selling the idea that our enemy is the leadership and their military forces, but the populace is our friend. With our advanced military weapons, we can defeat the defined "enemy" and then the populace will embrace us as liberators.
What I don't know is where this idea originated. My only guess was that it grew out of the reconstruction era in postwar Germany where civilian resistance was minimal and largely theoretical understandings of the Soviet domestic political climate.
Both of these seem naive. The Allies let the Germans starve for a couple of years after the war and most felt this was a better alternative than their experience with the Soviets. Despite Stalin and his repression, the Russians took massive losses and fought for the Soviet state. Much of this was compelled, but at the same time the populace did it.
Yet somehow, it's become a cornerstone of US military policy that the civilian population is at worst neutral and most likely supports US goals, not to mention US belief systems and values. Which is ironic if you look at most of the American and British propaganda from WWII, which sold the idea that the enemy nations were subhuman races which deserved to be wiped off the map.
This was my thinking.
It's been ages since I've used ipfw as a WAN simulator, but my memory of it is normally around a fairly static kind of configuration of latencies and bandwidth.
Simulating a cellular link that might hop between LTE and 1x kinds of data might be tough to do without some kind of engine which dynamically reprograms dummynets for vastly different bandwidth/latency scenarios to better simulate a node moving between 1x and LTE speeds. When I built a WAN simulator, I did to actually simulate known WAN link performance parameters.such as bandwidth and latency. I didn't have to worry about my link switching from multilink T1s to 512K frame relay to 56k dynamically.
They could have also provided a ton of statistical profile data so that the simulations closely mirrored real-world throughput associated with various media, especially common variability patterns.
A nice GUI front-end would be useful too, with actual throughput measured.
USB would not be desirable for internal system use, too much overhead. It is well designed for the purpose it has but you wouldn't want it for everything.
But what is "too much overhead" when the transport link gets fast enough? If USB4 ends up with 20 GBit/sec, overhead for anything but SAN shelf backplanes really won't matter.
I actually think I *would* want it for everything. One connector for disks and other peripherals, usable internally and externally. The way they package SSDs now you wouldn't even need to bother with an enclosure.
I think the real problem USB specifically has is a marginal performance history with USB2 devices (high CPU usage, low throughput) and Microsoft's steadfast refusal to allow Windows installs to USB devices, even USB3 (which makes no sense, really, when I can benchmark an ordinary PNY USB3 stick @ 110 MByte/sec read and 60 MByte/sec write).
If Windows could boot off USB3, eSATA would be largely forgotten as faster but with clunky, limited cabling and even SATA as a connector internal standard might get relegated to "enthusiast" boards where some minor performance boost was seen as valuable. M.SATA adoption would end up only in places where extreme miniaturization matters or the same enthusiast crowds.
I know it sounds crazy, but it sure seems a lot less crazy with USB3.1 supposed to hit 10 Gbits.