RAID-5 is simply not recommended for applications requiring performance, it's perfectly suited to archival storage (if you're not using tapes) and fileservers (where you would mostly see sequential read/write), and with a spare drive or two they're in most cases "safe enough", but when it comes to systems running 24/7 with loads ranging from "a lot" to "omfg" there is simply no alternative to a sweet-ass RAID-10-system.
Which probably isn't especially relevant to a desktop PC using fakeraid.;)
And I still say that you need a pretty unrealistic corner-case for a RAID5 to be slower than a single disk (averaged over a reasonable timeframe, so not just a single IOP.;) )
I'm certainly not going to argue RAID5 is as ast, or faster, than RAID10 (except maybe in long streaming writes). That would be dumb.
Not really. Any cases that force a write without a fully-cached block will initiate a read from all the disks and new parity calculation and then a write.
Actually, you would need to force writes that are smaller than a full stripe - and lots of them, constantly (to override the faster performance that will be had with more typical access patterns).
That would probably need to be coupled with an insanely braindead RAID5 implementation that didn't do any sort of block-level caching and/or write coalescing.
Personally, I'd call that a pretty malicious and unrealistic corner-case (and one that single drives are going to struggle badly with as well).
Show me an example of a government intervention in the economy which turned out well.
Is your benchmark for "turned out well" the most efficient allocation of money, or not having people dying of starvation because they can't afford to buy food ?
No need to get fancy with an "RSS feed". rTorrent, at least, can be configured to monitor a directory for.torrent files and automatically start downloading when one appears. You could set this up, then simply push out your.torrent file to each site with something like scp or rsync.
So if I follow your advice, I will have to keep a VGA monitor around to switch to to check to make sure I'm not being boned, as well as check the page design of every website I go to just to make sure their isn't a WMV video with DRM enabled.
No. If you follow my advice you will be able to see that your assertion the Protected Path is always active is false.
The Protected Path is only active when the playback application asks for it. That is fact. That is how the system was designed. It is trivially simple to verify by outputting some media to a device that is incapable of supporting the Protected Path.
Let me put it another way. If your assertion that the Protected Path is always active (and, by extension, that Vista is applying DRM controls to media that is not DRM-encumbered) were correct, then the only way to get video and audio out of a Vista box at all would be if the video and audio hardware was DRM-capable. Since, clearly, you can get video and audio out of a Vista machine without DRM-capable displays and sound systems, your assertion is wrong.
Wow, that is so much easier than just avoiding Vista like the clap! Thanks a lot!
"Avoiding Vista" won't help you in the slightest if you have DRM-encumbered media you wish to use. Conversely, not avoiding Vista will have zero impact on your ability to watch non-DRM-encumbered media.
Vista is irrelevant in the context of DRM, because DRM is an attribute of the media, not the tool used to play it. If you have media you can't use because of DRM, then your complaint is with the distributor of the media, not Vista. Vista is not stopping you from doing anything you would be able to do with a different platform.
But hey, if you won't believe me, maybe you will believe Microsoft [kareldonk.com] and the link that gives you their full answer to Gutmann's assessment of Vista is at the top of the page.
They start by saying Microsoft's response to Gutman "basically confirmed everything [he] wrote" . With such a blatant lie right at the start, it's pretty obvious the rest isn't going to be remotely objective.
And I quote MSFT "Will Windows Vista content protection features increase CPU resource consumption? Yes. [...]"
I don't recall disagreeing DRM would use a small amount of additional processing power when it was in use. Perhaps you have me confused with your straw man ?
So there it is, straight from the horses mouth. But believe what you will. But you can't seriously think we went from these [microsoft.com] system requirements to the incredible bloat of Vista(anything less than a dual core with 2Gb in my exp=thrashmonster) simply because of Aero and UAC, do you?
If you think the only difference betweem XP and Vista is "Aero and UAC", you're too stupid and ignorant to be arguing about it.
Filevault uses the same fragile and kludgy approach that Stacker, SuperStor, and others did a couple of decades ago. A big file that is loopback-mounted (I believe in 10.5 it's actually several smaller files, but the principle remains the same).
In contrast, in NTFS encryption or compression is a per-file attribute that can be disabled or enabled at will. ZFS isn't quite as nice, as it applies across the whole "drive", but it's still far, far more elegant and robust than a loopback-mounted file.
I'll bet that this whole thing really has little to do with "Christian nutters" as you so eloquently put it.
The whole 'censor the internet' thing has been a vote-grab by Labor to pander to the permanently-morally-outraged socially conervatives, who became increasingly more visible during the last decade.
It is exacerbated by the balance of power being held by "Christian nutters".
While I would never suggest the whole "corporate lobbying" thing doesn't happen in Australia, it is nowhere near as widespread and blatant as the US.
In short, yes, it is the "Christian nutters" who are responsible.
You see a new OS, you poke around the GUI. You explore the thing and
you use your ABSTRACT UNDERSTANDING of what you need to do to help you
figure out the details.
The average person doesn't have an ABSTRACT UNDERSTANDING of how an OS [interface] works, any more than they do about a car. Pull here, push there, turn that.
That anyone even considers (let alone carries out) "retraining" between Windows or Office releases is proof that the average person hasn't the vaguest level of "abstract understanding" when it comes to computers.
Windows != "standards". And, by the time a 7th grader enters the work force, Windows will be less like the XP he's using now than Mandriva is like XP.
Rubbish. There hasn't been a major change to the way the Windows GUI functions since Windows 95. Even Vista is immediately recognisable as a derivative.
To put that in context, the "Linux GUI" in the same timeframe has gone from fvwm95 to KDE 4.x. Heck, there's probably a bigger functional difference between KDE 3.0 and 4.0 than there is between Windows 95 and Vista.
Compared to every other remotely mainstream platform, the Windows GUI has been a rock of consistency (for over 13 years). The OS X GUI has probably changed more since 2000 than Windows' has since 1995.
Each new version of Microsoft Word is less like the previous version or Word than that previous version was to Star Office.
This is, similar to the above, utter garbage. Word 2008 is probably the first big change in the way the Word interface actually functions since the days of Windows 3.0.
RAID 5 is slower than single disks when writing, and little faster when reading.
What ? You'd need a pretty malicious (and unrealistic) corner-case for even writing to be slower on RAID5 vs single disk, and read performance will be similar to RAID0.
If you steal, don't be surprised if you're labeled a thief. If you support the right to steal, don't be surprised if you also receive a label. That's just telling it like it is.
At which point is killing a living organism not 'murder' ? Is this a standard you a
Nice thought, but the Windows Driver Foundation(WDF) that is built to support the protected path has to be able to check the entire path BEFORE you run ANY DRM videos, otherwise you could just pull an Alcohol 120% and install the "screw u" driver BEFORE playing the DRM and it's bye bye DRM.
No, you couldn't, because the system wouldn't be able to verify a protected path through the 'broken' driver and wouldn't work (to say nothing of driver signing not allowing it to install).
But don't believe me, read about it at Forbes [forbes.com] in an article written by the great Bruce Schneier [geekz.co.uk].
Schneir is basing his assessment from Peter Gutman's paper, the presumptions and errors of which have already been refuted numerous times.
The protected path is only activated when an application asks for it. This is fact. This is how the system is designed. If you aren't playing DRM-encumbered media (or, to be doubly sure, if you aren't using a DRM-capable application), then protected path is not enabled. Despite common FUD on Slashdot, Vista will not apply DRM controls to your DRM-free media (an *application* running on Vista might, but the OS will not).
It is so trivially simple to verify that the protected path is not always active (just output video to a VGA-only display, or audio via an SPDIF out), that it's astounding anyone still tries to argue otherwise.
Actually, no. When Tanenbaum was writing this, Windows NT was still in production, so he didn't have the full information about it.
I think it's pretty safe to say that in 1992, a year before release, the kernel architecture of Windows NT was well and truly known, especially to someone who would have a significant academic interest to it (and was probably running betas).
Windows NT is now commonly referred to as having a "hybrid kernel."
Yes. Hence the "-ish" I put on the end.
A "Hybrid Kernel" (like, say, Windows NT and OS X) is designed like a microkernel (so lots of small components communicating via well-known interfaces and message passing), but many of those bits actually run in kernel space (thus reducing the performance impact of the message passing).
It should be fairly obvious why a "Hybrid Kernel" has a more modular design than a monolithic kernel.
It's worth noting that, over time, the level of "microkernelness" in Windows NT has varied. NT 3.x, for example, ran the display system in user space. It was moved into kernel space for NT 4.x-5.x, but then moved back into user space for NT 6.x (Vista).
So Vista is slower on the same hardware? Big deal, every OS is.
Well, to be fair, OS X improved performance on the same hardware with every version up until 10.5.
Of course, the flipside to that is it was so unbelievably slow that they didn't really have anywhere to go but up.
At least you could buy an affordable machine that would run Vista well at release. It took years from OSX's first release before Apple had _any_ machine capable of running it well. No version of Windows has ever had performance that bad.
Uh...yeah. They are. You can run Leopard just fine on a 6 year old Mac just fine...why don't you try doing the same with Vista and a 6 year old PC, and get back to us.
If your assessment of OS X on a 6-year-old Mac is 'fine', then your assessment of Vista on an equivalent-costing 6-year-old PC would be 'great'.
it's only been six weeks since the u.s. election -- and already people are forgetting the importance of anonymity.
In no way did I suggest anonymity wasn't important.
it's kind of a shame that ms. dyson doesn't realize that.
I'm sure she does. She isn't saying anonymity isn't important, she's saying it shouldn't be encouraged [needlessly - presumably because of how people behave behind the veil of anonymity].
Use of anonymity when it is necessary, is critical to the proper functioning of free society. Using anonymity when it is unnecessary, just leads to people acting like arseholes.
Slashdot Mirror
RAID-5 is simply not recommended for applications requiring performance, it's perfectly suited to archival storage (if you're not using tapes) and fileservers (where you would mostly see sequential read/write), and with a spare drive or two they're in most cases "safe enough", but when it comes to systems running 24/7 with loads ranging from "a lot" to "omfg" there is simply no alternative to a sweet-ass RAID-10-system.
Which probably isn't especially relevant to a desktop PC using fakeraid. ;)
And I still say that you need a pretty unrealistic corner-case for a RAID5 to be slower than a single disk (averaged over a reasonable timeframe, so not just a single IOP. ;) )
I'm certainly not going to argue RAID5 is as ast, or faster, than RAID10 (except maybe in long streaming writes). That would be dumb.
Not really. Any cases that force a write without a fully-cached block will initiate a read from all the disks and new parity calculation and then a write.
Actually, you would need to force writes that are smaller than a full stripe - and lots of them, constantly (to override the faster performance that will be had with more typical access patterns).
That would probably need to be coupled with an insanely braindead RAID5 implementation that didn't do any sort of block-level caching and/or write coalescing.
Personally, I'd call that a pretty malicious and unrealistic corner-case (and one that single drives are going to struggle badly with as well).
Show me an example of a government intervention in the economy which turned out well.
Is your benchmark for "turned out well" the most efficient allocation of money, or not having people dying of starvation because they can't afford to buy food ?
No need to get fancy with an "RSS feed". rTorrent, at least, can be configured to monitor a directory for .torrent files and automatically start downloading when one appears. You could set this up, then simply push out your .torrent file to each site with something like scp or rsync.
So if I follow your advice, I will have to keep a VGA monitor around to switch to to check to make sure I'm not being boned, as well as check the page design of every website I go to just to make sure their isn't a WMV video with DRM enabled.
No. If you follow my advice you will be able to see that your assertion the Protected Path is always active is false.
The Protected Path is only active when the playback application asks for it. That is fact. That is how the system was designed. It is trivially simple to verify by outputting some media to a device that is incapable of supporting the Protected Path.
Let me put it another way. If your assertion that the Protected Path is always active (and, by extension, that Vista is applying DRM controls to media that is not DRM-encumbered) were correct, then the only way to get video and audio out of a Vista box at all would be if the video and audio hardware was DRM-capable. Since, clearly, you can get video and audio out of a Vista machine without DRM-capable displays and sound systems, your assertion is wrong.
Wow, that is so much easier than just avoiding Vista like the clap! Thanks a lot!
"Avoiding Vista" won't help you in the slightest if you have DRM-encumbered media you wish to use. Conversely, not avoiding Vista will have zero impact on your ability to watch non-DRM-encumbered media.
Vista is irrelevant in the context of DRM, because DRM is an attribute of the media, not the tool used to play it. If you have media you can't use because of DRM, then your complaint is with the distributor of the media, not Vista. Vista is not stopping you from doing anything you would be able to do with a different platform.
But hey, if you won't believe me, maybe you will believe Microsoft [kareldonk.com] and the link that gives you their full answer to Gutmann's assessment of Vista is at the top of the page.
They start by saying Microsoft's response to Gutman "basically confirmed everything [he] wrote" . With such a blatant lie right at the start, it's pretty obvious the rest isn't going to be remotely objective.
And I quote MSFT "Will Windows Vista content protection features increase CPU resource consumption? Yes. [...]"
I don't recall disagreeing DRM would use a small amount of additional processing power when it was in use. Perhaps you have me confused with your straw man ?
So there it is, straight from the horses mouth. But believe what you will. But you can't seriously think we went from these [microsoft.com] system requirements to the incredible bloat of Vista(anything less than a dual core with 2Gb in my exp=thrashmonster) simply because of Aero and UAC, do you?
If you think the only difference betweem XP and Vista is "Aero and UAC", you're too stupid and ignorant to be arguing about it.
Real world usage is about 250 for USB 2.
So groups of people can work together on one work.
Huh? I don't see what's so "low" about FileVault.
Filevault uses the same fragile and kludgy approach that Stacker, SuperStor, and others did a couple of decades ago. A big file that is loopback-mounted (I believe in 10.5 it's actually several smaller files, but the principle remains the same).
In contrast, in NTFS encryption or compression is a per-file attribute that can be disabled or enabled at will. ZFS isn't quite as nice, as it applies across the whole "drive", but it's still far, far more elegant and robust than a loopback-mounted file.
I'll bet that this whole thing really has little to do with "Christian nutters" as you so eloquently put it.
The whole 'censor the internet' thing has been a vote-grab by Labor to pander to the permanently-morally-outraged socially conervatives, who became increasingly more visible during the last decade.
It is exacerbated by the balance of power being held by "Christian nutters".
While I would never suggest the whole "corporate lobbying" thing doesn't happen in Australia, it is nowhere near as widespread and blatant as the US.
In short, yes, it is the "Christian nutters" who are responsible.
OS X could, with hardly any effort at all on Apple's part. FileVault implements an encrypted disk system.
Well, with the bar set that low then you can already create a compressed .dmg file, mount that and put all your data in it, tpday.
It hardly compares with the approaches of NTFS, or even ZFS, though.
If your data is not in your possession, how do you know others won't see it or edit it without your permission?
Encryption ?
Not to mention that all modern OSes can do file system compression by themselves nowadays...
What definition of "modern OS" are you using ?
Windows can, Solaris can, FreeBSD is a maybe.
OS X cannot. Linux cannot (at least not with any of its mainstream filesystems).
I'd love to see a block-based HSM device-mapper layer. Keep copies of frequently accessed blocks in flash, and migrate stuff in and out as needed.
ZFS will do that quite nicely.
You see a new OS, you poke around the GUI. You explore the thing and you use your ABSTRACT UNDERSTANDING of what you need to do to help you figure out the details.
The average person doesn't have an ABSTRACT UNDERSTANDING of how an OS [interface] works, any more than they do about a car. Pull here, push there, turn that.
That anyone even considers (let alone carries out) "retraining" between Windows or Office releases is proof that the average person hasn't the vaguest level of "abstract understanding" when it comes to computers.
Windows != "standards". And, by the time a 7th grader enters the work force, Windows will be less like the XP he's using now than Mandriva is like XP.
Rubbish. There hasn't been a major change to the way the Windows GUI functions since Windows 95. Even Vista is immediately recognisable as a derivative.
To put that in context, the "Linux GUI" in the same timeframe has gone from fvwm95 to KDE 4.x. Heck, there's probably a bigger functional difference between KDE 3.0 and 4.0 than there is between Windows 95 and Vista.
Compared to every other remotely mainstream platform, the Windows GUI has been a rock of consistency (for over 13 years). The OS X GUI has probably changed more since 2000 than Windows' has since 1995.
Each new version of Microsoft Word is less like the previous version or Word than that previous version was to Star Office.
This is, similar to the above, utter garbage. Word 2008 is probably the first big change in the way the Word interface actually functions since the days of Windows 3.0.
RAID 5 is slower than single disks when writing, and little faster when reading.
What ? You'd need a pretty malicious (and unrealistic) corner-case for even writing to be slower on RAID5 vs single disk, and read performance will be similar to RAID0.
If you steal, don't be surprised if you're labeled a thief. If you support the right to steal, don't be surprised if you also receive a label. That's just telling it like it is.
At which point is killing a living organism not 'murder' ? Is this a standard you a
Nice thought, but the Windows Driver Foundation(WDF) that is built to support the protected path has to be able to check the entire path BEFORE you run ANY DRM videos, otherwise you could just pull an Alcohol 120% and install the "screw u" driver BEFORE playing the DRM and it's bye bye DRM.
No, you couldn't, because the system wouldn't be able to verify a protected path through the 'broken' driver and wouldn't work (to say nothing of driver signing not allowing it to install).
But don't believe me, read about it at Forbes [forbes.com] in an article written by the great Bruce Schneier [geekz.co.uk].
Schneir is basing his assessment from Peter Gutman's paper, the presumptions and errors of which have already been refuted numerous times.
The protected path is only activated when an application asks for it. This is fact. This is how the system is designed. If you aren't playing DRM-encumbered media (or, to be doubly sure, if you aren't using a DRM-capable application), then protected path is not enabled. Despite common FUD on Slashdot, Vista will not apply DRM controls to your DRM-free media (an *application* running on Vista might, but the OS will not).
It is so trivially simple to verify that the protected path is not always active (just output video to a VGA-only display, or audio via an SPDIF out), that it's astounding anyone still tries to argue otherwise.
Actually, no. When Tanenbaum was writing this, Windows NT was still in production, so he didn't have the full information about it.
I think it's pretty safe to say that in 1992, a year before release, the kernel architecture of Windows NT was well and truly known, especially to someone who would have a significant academic interest to it (and was probably running betas).
Windows NT is now commonly referred to as having a "hybrid kernel."
Yes. Hence the "-ish" I put on the end.
A "Hybrid Kernel" (like, say, Windows NT and OS X) is designed like a microkernel (so lots of small components communicating via well-known interfaces and message passing), but many of those bits actually run in kernel space (thus reducing the performance impact of the message passing).
It should be fairly obvious why a "Hybrid Kernel" has a more modular design than a monolithic kernel.
It's worth noting that, over time, the level of "microkernelness" in Windows NT has varied. NT 3.x, for example, ran the display system in user space. It was moved into kernel space for NT 4.x-5.x, but then moved back into user space for NT 6.x (Vista).
So Vista is slower on the same hardware? Big deal, every OS is.
Well, to be fair, OS X improved performance on the same hardware with every version up until 10.5.
Of course, the flipside to that is it was so unbelievably slow that they didn't really have anywhere to go but up.
At least you could buy an affordable machine that would run Vista well at release. It took years from OSX's first release before Apple had _any_ machine capable of running it well. No version of Windows has ever had performance that bad.
I'm typing this from OS X Lepoard on my 12" PowerBook G4.
And ? Vista can be run usably on PCs that were around years before your PowerBook was even made.
Uh...yeah. They are. You can run Leopard just fine on a 6 year old Mac just fine...why don't you try doing the same with Vista and a 6 year old PC, and get back to us.
If your assessment of OS X on a 6-year-old Mac is 'fine', then your assessment of Vista on an equivalent-costing 6-year-old PC would be 'great'.
So does anyone here have Win7 running and is the protected path DRM on?
The DRM is only used when you're playing DRM-encumbered media. No DRM-encumbered media, no protected path active, no (insignificant, anyway) slowdown.
Of course it's not. Well, assuming you're talking about the Windows kernel anyway. http://oreilly.com/catalog/opensources/book/appa.html [oreilly.com]
What ? As Tanenbaum points out, Windows NT is a microkernel (-ish) OS. That's about as 'modular' as it gets, from the perspective of the kernel.
Anonymity is important because it gives people the power to say un-popular things that need to be said without getting mob-lynched for it.
I don't think anyone is disagreeing with this.
However, people abuse anonymity too. People act *much* differently when being anonymous, and it's usually not for the better.
Which is, I believe, the reasoning behind the suggestion that it isn't something that should be encouraged.
it's only been six weeks since the u.s. election -- and already people are forgetting the importance of anonymity.
In no way did I suggest anonymity wasn't important.
it's kind of a shame that ms. dyson doesn't realize that.
I'm sure she does. She isn't saying anonymity isn't important, she's saying it shouldn't be encouraged [needlessly - presumably because of how people behave behind the veil of anonymity].
Use of anonymity when it is necessary, is critical to the proper functioning of free society. Using anonymity when it is unnecessary, just leads to people acting like arseholes.