Recovering data from a partially failed disk is doable. Recovering data from a totally failed disk is not, unless you're willing to pay the nice folks at a data recovery service lots of money. Recovering data from a stripeset with a failed disk is more in the latter category than the former. RAID does affect how hard it is to recover data.
Hardware RAID controllers are wonderful, and they buy you roughly nothing for RAID0. Without parity, there's no need for the caching that a RAID controller has, and the overhead of maintaining a stripe volume is miniscule--roughly 2%.
MTBF is a remarkably vague number for hard drives, because the standard deviation is so wide. It's also measured in a very specific manner, which bears little resemblance to the real world. Practically speaking, you're typically going to get between three and five years out of a hard drive before it starts to error out.
Now if you're actually backing stuff up, then by all means--use RAID0. Blow a hundred bucks on a RAID controller that's not being used, if it makes you happy. But please try to relax a bit, and not spit bile at people just for pointing out that RAID0 decreases a system's reliability.
I think a lot of people are confused here. There are three basic types of desktop machines.
1) An office PC. Runs some office package, web browser, acrobat reader, etc. a P3/800MHz with a low-end graphics card is FINE for this. 2) A gaming machine. You all know what this is. 3) A workstation. This is NOT the office PC. This is a serious machine for serious work--CAD, 3D modelling, number crunching, etc.
The requirements for a workstation and a gaming machine are similar but not identical. A workstation may have slower graphics, but accurate. No fudging or edge-blending to make things look prettier (or run faster) at the expense of mathematically correct representation. A workstation also is likely to have multiple processors, since they can be fully used by most software packages one would be looking at.
Hmm. Take a step back for a second. Your recommendations are based on your weaknesses. Recommending Windows/MacOS because you can't help mucking about in a Unix/Linux environment isn't entirely applicable. (understandable, though:-)
At any rate, this wasn't a gaming/office machine that they were building, it was a gaming/workstation. In my world (and I believe the world of the article's authors), a workstation is a seriously robust, high-end, number-cruncher. If you need a workstation and want to play games, then build a workstation and it'll probably play games.
No doubt. I went to a walk-in clinic because of a horrible pain running from my mid-back to my jaw. A 30-second 'examination' said that I was sitting badly at work, I should adjust my chair and take these muscle relaxants. End of appointment.
Three days later, I was in emergency, unable to close my mouth. I had an abcessed tooth, which had lead to my back problems. I then got put on the RIGHT pills (antibiotics--I'm not a fan, but in certain cases they're absolutely necessary), and went for dental surgery later.
The problem is twofold: First, many (most?) doctors don't actually examine the patient properly--they take the patient-reported symptoms at face value and treat those, rather than investigating. Seondly, they get most of their drug information from the drug manufacturers who are myopic at best.
If you want to find out what a drug REALLY does, ask your pharmacist when you get a prescription filled--they're required to know far more about what they're handing out than the doctors are, and I've generally found them much happier to share information.
1) Calendar stuff. Could be done on paper. 2) Making reminder/shopping lists. Could be done on paper. 3) Recording service calls. Could be done on paper, but painfully. 4) Playing iRogue. 5) using it as a serial terminal for broken hardware.
#5 is the reason I got the thing, and it's absolutely irreplaceable in that aspect. It makes the idea of dragging around a laptop seem like the dark ages. (although to be fair, my laptop IS from the dark ages) It's tiny--I can carry a terminal and cables around in my pockets, and never worry about batteries dying halfway through a session. Admittedly this isn't a big deal for most people, but for me it's critical.
#4 is essential.
#1-3 all could be done with a notepad, but for the amount of stuff I put into my Palm Vx, I'd be carrying a LOT of paper around! One thing that it's very nice for, is building a form to enter stuff into.
For a few things, I actually need one. For the other bits, it's a lot handier for me, and that's worth paying money for.
OK, technically you're right. You can't prove a negative claim.
But Occam's razor, the flat out negation by everyone involved, and common sense ALL agree that it just isn't rational. Microsoft could easily give SCO $50M publicly and state that it was to help further their IP claims against IBM, without suffering. They don't NEED to futz around with cloak-and-dagger funding. Furthermore, it would be much more damaging to them to do something like this, and then be found out afterwards. How would they be found out? They're a public company--look at their financials, which are a matter of public record.
So the only way this could happen is if a vast conpiracy took place to illegally cook the books at Microsoft, in order to achieve something they could safely do in the public eye.
Thus I say, that to within a reasonable degree of doubt, it just isn't happening.
As an aside, the subject of your.sig is one of the biggest and most poorly researched conspiracy theories I've seen. It's nice to have a Nobel Laureate that you can quote on the subject, but it doesn't help when he's out of his league and half demented. Have you ever met Kary Mullis? I used to work for a company that collaborated with him, and the guy is a certifiable nutcase. If you're going to quote his opinions on HIV/AIDS, then you should also quote his theories on alien seeding.
There's a thing you have to understand about the stock market, and all it stands for.
A company is worth exactly what someone is willing to pay for it. If a public company had $50 billion in net assets (i.e. cash and real products, subtracting outstanding debt), and had 500 million shares trading publically for $20/share, then their total worth isn't $50 billion--it's $10 billion. It doesn't matter what they make, what their future is, or how much cash they have in their pockets. Unless they go bankrupt, the only valuation is the stock market. It sometimes sucks and often doesn't reflect a company's inherent worthiness, but that's also why the stock market moves.
Also, keep in mind that Sun has a lot more stock out there than SCO. Even with the stupid devaluation of Sun stock, they have far more market value than SCO.
OK, this is about as stupid as anything/. has EVER posted. It's pure conspiracy theory, and has been flatly refuted by everyone involved. Hell, the eWeek article only gets by at all by saying, "some in the open source community suspect..."
First of all, let's quit ragging on the MBAs. The vast majority of them have souls, and do useful work. In fact, a friend of mine is an MBA who advises venture capitalists on what technology not to waste their money on. He's was Civil Engineer with his PEng and nearly ten years experience, before going back for his MBA.
Now, on to the matter at hand. The libertarian ideal of an underground internet. Lovely, wonderful, and utterly unlikely. There will be _some_ people who build something like it, but they'll be a tiny number compared to the internet population.
Understand that people are fundamentally lazy, and have a nearly infinite capacity for pain delivered in slowly increasing increments. Muck with redirects now, then withdraw them, then reinstate them in certain rare cases. Increase them gradually, and in a year nobody will have noticed that every typo they make sends them to advertising.
Sun is on the Linux bandwagon--this week. In fact, we had a presentation yesterday where they actually had a viable Linux solution for the desktop. That's right, we got to see Mad Hatter, and it's darned pretty.
Linux on the high-end Sparc has never been a good idea from a technical point of view. Solaris and Sparc sell each other. People want Solaris, they buy Sparc hardware. They want Sparc hardware, they don't want to run anything but Solaris on it. Outside the realm of the desktop land, this is how things are. Part of Sun's problem is that they've always been a technology company, and when their technology gets behind they start playing with marketing, which they suck at.
If Sun can stick to their guns now--offer (and really PUSH) Linux on their x86 stuff, offer (again, REALLY REALLY PUSH) their Linux packages to companies for the desktop, and then quit talking about Linux on the high end--they might get across the message that they're actually "with it" (i.e. supporting Linux) without trashing their core business. Of course to do that, they're going to have to get the UltraSparc IV out on schedule and with a faster core than predicted. If there's an UltraIV running at 2.5GHz ON THE MARKET by the end of 2004, then their hardware can survive.
There's one thing that Sun needs to do to change their image: Maintain consistency! Actually work up a GOOD strategy, and then stick with it for more than two weeks running.
"I'm just refuting Gates' claims that MS patches faster than "Linux"."
Heh. Now past behaviour is certainly an indicator--I'll agree with that. The counter-argument is that four years is a LONG time in computer history (back in the NT4.0 days), and also that Microsoft has (supposedly) made a very deliberate, explicit, and massive change in their whole security patch policy (procedures, change management, etc.).
Now I'm not suggesting that MS really IS faster than Linux at patching, but I really don't think that data from nearly half-a-decade ago can be considered relevant anymore, even as much of a trend indicator.
Hmm. Interesting. However, if the goal of using the GPL is to invoke change, but that selfsame change discourages people from using your software, then what have you won? It's something like building a perfect village and then having aggressive rules to enforce that perfection. Would people move there? Is it still utopia if it has no citizens?
This is a very interesting point. The question remains however, about what happens when nobody uses the free software, because of the license that keeps it free?
I know, people use Linux. A lot. However, his point suggests that the FSF might cause Linux to remain a hobbiest OS, by and large.
1) Technological solutions only solve purely technological problems. 2) There are no purely technological problems.
Spam is a societal problem. Spammers refuse to acknowledge that they're stealing and committing fraud. Filtering is a technological 'solution' to spam, and not even a good one at that. First of all, it doesn't stop anything. If the spammers can get out a million messages and one is responded to, then they're happy. If 900,000 of those messages are blocked, then they'll send out ten million or half a billion, to get their one response. There's no incremental cost to the spammer by filtering on the client/ISP side. All this leads to is more active spammers, fighting hard to defeat the filters or in other words--an arms race.
A better bit of technology would be greylisting, because the worst case scenario is that it drives the incremental cost up for the spammers. However, it's still not a solution. The solution lies in pressing serious charges against the fraudsters and thieves who spam. We don't need special anti-spam laws (which would force spam into becoming a free speech issue), but we just need to enforce the laws we have.
In this particular case, there are no freedom of speech issues anyways--the bloody spammers started the whole court case in the first place!
The ONLY way to stop spam outright is to stop the spammers from sending it in the first place; and the only way to do that is by social and legal means, NOT technological. (unless you're counting handcuffs:-)
"True, but its the data we have, unless you know of a more recent study."
The fact that it's the best available doesn't automatically make it relevant data. What you're saying here boils down to this: Bill Gates is lying or wrong, because what he says his company does today wasn't the case four years ago.
Honestly, that's totally nonsensical. You should know better, Crispin.
James Dyson once said something to the effect of, "people imagine the designer as the guy in a salmon-coloured shirt who comes in at the end and says 'put fins on it.' It's not like that--good products come from good design, which goes hand in hand with good engineering from the ground up."
The more I hear about style as a separate field, or as something different from form function and execution, the more I think that someone has missed the point. Style can't be separated from the product itself. Talking about it in isolation, or treating it as though it were different from any other aspect of creation is silly.
"unfortunately, it is NOT an essential part of the market."
Well, maybe essential isn't the world I should have used. How about inextricable? There's nothing about the market that inherently requires OSS, but there _ARE_ several chunks of the market that currently rely on actual OSS software. In that sense, we can't (and shouldn't) get rid of OSS, nor can (or should) we get rid of closed source software.
"Understandable, and will you keep the same attitude the next time Microsoft releases some sort of patch for their operating system?"
Yes, but with a caveat. I object to the fundamental design of Windows, wherein much of the nonessential stuff is put into the kernel, and results in a weaker OS by design. I also object to them constantly paying lip service to 'secure coding practices, revolution, etc.' when they're not even keeping up to the rest of the market (open OR closed source).
But I will always applaud tested patches being released by companies in a timely fashion.
That's all there is to it. Infrastructure has to be neutral. That means that HTML must be a platform-neutral standard, DNS resolution must be vendor-neutral, etc., etc.. If your browser doesn't want to follow standards, or if it wants to go to its own page AFTER getting a neutral (Error 404) reply, then so be it. But implement it at the browser level, NOT the infrastructure level.
Please, "Linux junkies" or better still, "OSS junkies." Most Unix professionals understand that OSS is neither a holy grail or guarantee of perfect software everytime.
Open software is an essential part of the market, but it's not magic. Bad programmers will still write bad code, and lazy reviewers will still miss bugs.
I'm a big fan of NAT, especially whne properly set up--as we know and as he acknowledges, there's nothing inherent to NAT that breaks the peer-peer model that works so well.
However, what I think he's objecting to comes down to ISP-level firewalls, out of control of the end user. I won't stand for that, but I'm afraid that he's right--it's likely to happen, and most people won't even notice it.
Slashdot Mirror
Heh. Just poking fun. Here you go:
privet.
Man, did you forget your meds today?
A few rebuttals to your comments here.
Recovering data from a partially failed disk is doable. Recovering data from a totally failed disk is not, unless you're willing to pay the nice folks at a data recovery service lots of money. Recovering data from a stripeset with a failed disk is more in the latter category than the former. RAID does affect how hard it is to recover data.
Hardware RAID controllers are wonderful, and they buy you roughly nothing for RAID0. Without parity, there's no need for the caching that a RAID controller has, and the overhead of maintaining a stripe volume is miniscule--roughly 2%.
MTBF is a remarkably vague number for hard drives, because the standard deviation is so wide. It's also measured in a very specific manner, which bears little resemblance to the real world. Practically speaking, you're typically going to get between three and five years out of a hard drive before it starts to error out.
Now if you're actually backing stuff up, then by all means--use RAID0. Blow a hundred bucks on a RAID controller that's not being used, if it makes you happy. But please try to relax a bit, and not spit bile at people just for pointing out that RAID0 decreases a system's reliability.
What, you think you're special just because you maintain your hedge?
I think a lot of people are confused here. There are three basic types of desktop machines.
1) An office PC. Runs some office package, web browser, acrobat reader, etc. a P3/800MHz with a low-end graphics card is FINE for this.
2) A gaming machine. You all know what this is.
3) A workstation. This is NOT the office PC. This is a serious machine for serious work--CAD, 3D modelling, number crunching, etc.
The requirements for a workstation and a gaming machine are similar but not identical. A workstation may have slower graphics, but accurate. No fudging or edge-blending to make things look prettier (or run faster) at the expense of mathematically correct representation. A workstation also is likely to have multiple processors, since they can be fully used by most software packages one would be looking at.
Hmm. Take a step back for a second. Your recommendations are based on your weaknesses. Recommending Windows/MacOS because you can't help mucking about in a Unix/Linux environment isn't entirely applicable. (understandable, though :-)
At any rate, this wasn't a gaming/office machine that they were building, it was a gaming/workstation. In my world (and I believe the world of the article's authors), a workstation is a seriously robust, high-end, number-cruncher. If you need a workstation and want to play games, then build a workstation and it'll probably play games.
No doubt. I went to a walk-in clinic because of a horrible pain running from my mid-back to my jaw. A 30-second 'examination' said that I was sitting badly at work, I should adjust my chair and take these muscle relaxants. End of appointment.
Three days later, I was in emergency, unable to close my mouth. I had an abcessed tooth, which had lead to my back problems. I then got put on the RIGHT pills (antibiotics--I'm not a fan, but in certain cases they're absolutely necessary), and went for dental surgery later.
The problem is twofold: First, many (most?) doctors don't actually examine the patient properly--they take the patient-reported symptoms at face value and treat those, rather than investigating. Seondly, they get most of their drug information from the drug manufacturers who are myopic at best.
If you want to find out what a drug REALLY does, ask your pharmacist when you get a prescription filled--they're required to know far more about what they're handing out than the doctors are, and I've generally found them much happier to share information.
Here's what I use my PDA for:
1) Calendar stuff. Could be done on paper.
2) Making reminder/shopping lists. Could be done on paper.
3) Recording service calls. Could be done on paper, but painfully.
4) Playing iRogue.
5) using it as a serial terminal for broken hardware.
#5 is the reason I got the thing, and it's absolutely irreplaceable in that aspect. It makes the idea of dragging around a laptop seem like the dark ages. (although to be fair, my laptop IS from the dark ages) It's tiny--I can carry a terminal and cables around in my pockets, and never worry about batteries dying halfway through a session. Admittedly this isn't a big deal for most people, but for me it's critical.
#4 is essential.
#1-3 all could be done with a notepad, but for the amount of stuff I put into my Palm Vx, I'd be carrying a LOT of paper around! One thing that it's very nice for, is building a form to enter stuff into.
For a few things, I actually need one. For the other bits, it's a lot handier for me, and that's worth paying money for.
yep, that's about right. That's what Bruce always says.
The thing is, he's right. And he's determined to get his point across, so he's going to keep saying it until people start listening.
OK, technically you're right. You can't prove a negative claim.
.sig is one of the biggest and most poorly researched conspiracy theories I've seen. It's nice to have a Nobel Laureate that you can quote on the subject, but it doesn't help when he's out of his league and half demented. Have you ever met Kary Mullis? I used to work for a company that collaborated with him, and the guy is a certifiable nutcase. If you're going to quote his opinions on HIV/AIDS, then you should also quote his theories on alien seeding.
But Occam's razor, the flat out negation by everyone involved, and common sense ALL agree that it just isn't rational. Microsoft could easily give SCO $50M publicly and state that it was to help further their IP claims against IBM, without suffering. They don't NEED to futz around with cloak-and-dagger funding. Furthermore, it would be much more damaging to them to do something like this, and then be found out afterwards. How would they be found out? They're a public company--look at their financials, which are a matter of public record.
So the only way this could happen is if a vast conpiracy took place to illegally cook the books at Microsoft, in order to achieve something they could safely do in the public eye.
Thus I say, that to within a reasonable degree of doubt, it just isn't happening.
As an aside, the subject of your
There's a thing you have to understand about the stock market, and all it stands for.
A company is worth exactly what someone is willing to pay for it. If a public company had $50 billion in net assets (i.e. cash and real products, subtracting outstanding debt), and had 500 million shares trading publically for $20/share, then their total worth isn't $50 billion--it's $10 billion. It doesn't matter what they make, what their future is, or how much cash they have in their pockets. Unless they go bankrupt, the only valuation is the stock market. It sometimes sucks and often doesn't reflect a company's inherent worthiness, but that's also why the stock market moves.
Also, keep in mind that Sun has a lot more stock out there than SCO. Even with the stupid devaluation of Sun stock, they have far more market value than SCO.
OK, this is about as stupid as anything /. has EVER posted. It's pure conspiracy theory, and has been flatly refuted by everyone involved. Hell, the eWeek article only gets by at all by saying, "some in the open source community suspect..."
It's not happening. Get over it.
First of all, let's quit ragging on the MBAs. The vast majority of them have souls, and do useful work. In fact, a friend of mine is an MBA who advises venture capitalists on what technology not to waste their money on. He's was Civil Engineer with his PEng and nearly ten years experience, before going back for his MBA.
Now, on to the matter at hand. The libertarian ideal of an underground internet. Lovely, wonderful, and utterly unlikely. There will be _some_ people who build something like it, but they'll be a tiny number compared to the internet population.
Understand that people are fundamentally lazy, and have a nearly infinite capacity for pain delivered in slowly increasing increments. Muck with redirects now, then withdraw them, then reinstate them in certain rare cases. Increase them gradually, and in a year nobody will have noticed that every typo they make sends them to advertising.
It won't work. Not enough to make any difference.
Ah. You may be an MBA, but are you an MBA bean counter?
Maybe you're exempt.
Sun is on the Linux bandwagon--this week. In fact, we had a presentation yesterday where they actually had a viable Linux solution for the desktop. That's right, we got to see Mad Hatter, and it's darned pretty.
Linux on the high-end Sparc has never been a good idea from a technical point of view. Solaris and Sparc sell each other. People want Solaris, they buy Sparc hardware. They want Sparc hardware, they don't want to run anything but Solaris on it. Outside the realm of the desktop land, this is how things are. Part of Sun's problem is that they've always been a technology company, and when their technology gets behind they start playing with marketing, which they suck at.
If Sun can stick to their guns now--offer (and really PUSH) Linux on their x86 stuff, offer (again, REALLY REALLY PUSH) their Linux packages to companies for the desktop, and then quit talking about Linux on the high end--they might get across the message that they're actually "with it" (i.e. supporting Linux) without trashing their core business. Of course to do that, they're going to have to get the UltraSparc IV out on schedule and with a faster core than predicted. If there's an UltraIV running at 2.5GHz ON THE MARKET by the end of 2004, then their hardware can survive.
There's one thing that Sun needs to do to change their image: Maintain consistency! Actually work up a GOOD strategy, and then stick with it for more than two weeks running.
Actually, you did say that. :-)
"I'm just refuting Gates' claims that MS patches faster than "Linux"."
Heh. Now past behaviour is certainly an indicator--I'll agree with that. The counter-argument is that four years is a LONG time in computer history (back in the NT4.0 days), and also that Microsoft has (supposedly) made a very deliberate, explicit, and massive change in their whole security patch policy (procedures, change management, etc.).
Now I'm not suggesting that MS really IS faster than Linux at patching, but I really don't think that data from nearly half-a-decade ago can be considered relevant anymore, even as much of a trend indicator.
Hmm. Interesting. However, if the goal of using the GPL is to invoke change, but that selfsame change discourages people from using your software, then what have you won? It's something like building a perfect village and then having aggressive rules to enforce that perfection. Would people move there? Is it still utopia if it has no citizens?
This is a very interesting point. The question remains however, about what happens when nobody uses the free software, because of the license that keeps it free?
I know, people use Linux. A lot. However, his point suggests that the FSF might cause Linux to remain a hobbiest OS, by and large.
Let's go over this again:
:-)
1) Technological solutions only solve purely technological problems.
2) There are no purely technological problems.
Spam is a societal problem. Spammers refuse to acknowledge that they're stealing and committing fraud. Filtering is a technological 'solution' to spam, and not even a good one at that. First of all, it doesn't stop anything. If the spammers can get out a million messages and one is responded to, then they're happy. If 900,000 of those messages are blocked, then they'll send out ten million or half a billion, to get their one response. There's no incremental cost to the spammer by filtering on the client/ISP side. All this leads to is more active spammers, fighting hard to defeat the filters or in other words--an arms race.
A better bit of technology would be greylisting, because the worst case scenario is that it drives the incremental cost up for the spammers. However, it's still not a solution. The solution lies in pressing serious charges against the fraudsters and thieves who spam. We don't need special anti-spam laws (which would force spam into becoming a free speech issue), but we just need to enforce the laws we have.
In this particular case, there are no freedom of speech issues anyways--the bloody spammers started the whole court case in the first place!
The ONLY way to stop spam outright is to stop the spammers from sending it in the first place; and the only way to do that is by social and legal means, NOT technological. (unless you're counting handcuffs
"True, but its the data we have, unless you know of a more recent study."
The fact that it's the best available doesn't automatically make it relevant data. What you're saying here boils down to this: Bill Gates is lying or wrong, because what he says his company does today wasn't the case four years ago.
Honestly, that's totally nonsensical. You should know better, Crispin.
James Dyson once said something to the effect of, "people imagine the designer as the guy in a salmon-coloured shirt who comes in at the end and says 'put fins on it.' It's not like that--good products come from good design, which goes hand in hand with good engineering from the ground up."
The more I hear about style as a separate field, or as something different from form function and execution, the more I think that someone has missed the point. Style can't be separated from the product itself. Talking about it in isolation, or treating it as though it were different from any other aspect of creation is silly.
"unfortunately, it is NOT an essential part of the market."
Well, maybe essential isn't the world I should have used. How about inextricable? There's nothing about the market that inherently requires OSS, but there _ARE_ several chunks of the market that currently rely on actual OSS software. In that sense, we can't (and shouldn't) get rid of OSS, nor can (or should) we get rid of closed source software.
"Understandable, and will you keep the same attitude the next time Microsoft releases some sort of patch for their operating system?"
Yes, but with a caveat. I object to the fundamental design of Windows, wherein much of the nonessential stuff is put into the kernel, and results in a weaker OS by design. I also object to them constantly paying lip service to 'secure coding practices, revolution, etc.' when they're not even keeping up to the rest of the market (open OR closed source).
But I will always applaud tested patches being released by companies in a timely fashion.
That's all there is to it. Infrastructure has to be neutral. That means that HTML must be a platform-neutral standard, DNS resolution must be vendor-neutral, etc., etc.. If your browser doesn't want to follow standards, or if it wants to go to its own page AFTER getting a neutral (Error 404) reply, then so be it. But implement it at the browser level, NOT the infrastructure level.
Please, "Linux junkies" or better still, "OSS junkies." Most Unix professionals understand that OSS is neither a holy grail or guarantee of perfect software everytime.
Open software is an essential part of the market, but it's not magic. Bad programmers will still write bad code, and lazy reviewers will still miss bugs.
Step one: Move away from the US. Consumer broadband in the US is lagging behind Europe, Australia, and (especially) Canada by a long ways.
I'm a big fan of NAT, especially whne properly set up--as we know and as he acknowledges, there's nothing inherent to NAT that breaks the peer-peer model that works so well.
However, what I think he's objecting to comes down to ISP-level firewalls, out of control of the end user. I won't stand for that, but I'm afraid that he's right--it's likely to happen, and most people won't even notice it.