Re:solaris8 is BS. only linux is enterprise ready
on
Free Solaris 8
·
· Score: 1
Well, troll or not, it's just too much fun.
it runs on crappy, unscalable hardware.
If you have any of this, since it's so crappy, I'll happily take it off your hands. I'll even give you an enterprise-class (by YOUR definition) peecee for it.
it is not very efficient with its SMP code
Nothing is, sadly. It's a hard problem.
lets compare CDE to GNOME
Ok, lets. CDE: sucks 1e56 Ll. GNOME: sucks 1e56 Ll. Hmmm...I think I'll take "none of the above," please.
if you want a real enterprise os, buy redhat.
Define "enterprise." Red Hat is great for...well, it's ok for...ok, well, Red Hat is a big company. They stamp a lot of CDs. And the red hats are kind of cool. What were we talking about again?
Guess again. I've used Linux since '93. I've used Solaris since '96. I'm directly responsible for (not mine) systems running both OS's, and a few others as well. Except for Windows, Solaris is my least favorite. I didn't order Solaris {2.,5.,}7 because I know better. So you're not much of a profiler. But let's continue.
Linux was modeled after Unix why again?? And Sun being the largest Unix distributor means??
Not a damn thing. Witness Microsoft. Lots o' market share, no product.
"Better to be thought an idiot that open your mouth and remove all doubt."
I know this isn't a support forum, but I do know such hardware pretty well. A few ideas: it may be that the machine is set to talk on serial ports instead of the GX. Try hooking up a terminal, 9600,8n1, and disconnecting the keyboard. If it works ok, see what input-device and output-device are set to ("printenv" at the prom prompt). If it's ttyX, you've found the problem. Do "setenv input-device keyboard" and "setenv output-device screen", plug the keyboard back in, reboot, and voila. Instant workstation. Or, try swapping out the GX for something else - even if you just have to borrow it. If the GX is truly fried, there's nothing you'll be able to do about it of course, so there's really nothing to do once you've found the source of the problem.
Solaris is a very good operating system on its native hardware (SPARC et al) it is however a little weaker on non-native hardware.
That's odd, you know; what I think of when comparing the two is Solaris and Linux on an identical midrange SPARC machine. An Ultra 2, 60, or 80, or a 450-ish machine. Of course, a comparison between the two on an identical peecee comes out about the same - Solaris loses. I don't see Solaris x86's shortcomings as being due to non-native hardware, but rather to inferior hardware, and Linux, to some degree, suffers from the same on that platform.
But on a real system, designed and crafted by men and women who know and love their trade, a system like one of the above, the peecee hardware barriers evaporate and the gloves come off. And the brilliant Sun engineers who designed and built those machines find out that their OS counterparts let them down. Badly. Have you ever run Linux on a SPARC? Even a low-end older system will work smashingly. But don't take this to mean that Linux can't take advantage of newer, faster hardware - it most certainly can. Unless you're one of a handful who need an E10k with dozens of processors, Linux is the only way to go on SPARCs. Try it; you'll never go back.
Don't take this the wrong way - Linux has its own problems. But those problems get fixed faster, and with better overall results, than the problems in Solaris. Is Linux the One and Only Solution? Not yet. But I can't find one instance where Solaris is really the best of breed. At the low end, lightweight products like QNX rule. In the low-to-midrange, Linux is king. And at the high end, IBM and SGI duke it out for the best massively scalable feature set. If you amalgamated the very best of all these into one product and called it Linux (for lack of a better name), I doubt you'd find any Solaris in it at all. But that's what you'll get, 10 years from now. And Solaris? A distant memory.
And of course, every year Linux squeezes more away from traditional vendors, who are left gasping for air at the summit of the high end, wondering whether this will be the year they finally have to throw in the towel and admit that they no longer make a product that people aren't running Linux on. Some, like SGI and IBM, have long since seen this coming, and realized that an OS is an expensive and unprofitable product to make, and taken steps to ensure the survival of their real business, hardware. At the same time they get to make sure that the best pieces of their system are incorporated into Linux so that their customers can continue to enjoy the very best. Maybe Sun just realizes they haven't anything to offer to the effort. After all, haven't they just given up on making money from Solaris? And yet they still refuse to unite behind the one product everyone else is agreeing on. There's a name for this: overweening pride.
Well, all I can say is that I'm going to miss Sun's hardware. When they finally follow the fading glory that was once SunOS into the great bitbucket in the sky, I can only hope the hardware teams land on their feet.
Re:What are the Differences?
on
Free Solaris 8
·
· Score: 1
WARNING: Clue included!
SunOS was BSD-based. Solaris is nearly pure System V. Have you ever actually used a BSD system?
My personal opinion of Solaris man pages is that they suck ass. Way, way too much space taken by tables with "SUNW,ZOIurgewqr" in them. Of course, since Solaris is seriously feature-starved, the man pages don't really have much to document, so they've got to fill them with something. But, again, that's just me. Now IRIX, there are some decent man pages.
To be fair, an E10k is an awfully nice machine. Linux would run on that too, if anybody could afford one. Your customer (and you!) can just be thankful that their application is suitable for clustering. Otherwise, the iron is essential. Operating system has little to do with it.
Big deal. More Slashdot non-news. Solaris is a second-rate operating system at best, and if we ever see the source it'll be under the not-even-close-to-Free-but-we're-muddying-the-wate rs SCSL. Not that the source would be worth much anyway. Probably better that we don't get it, lest some fools try to pollute good operating systems with it.
The only things Sun sells that anyone would want are made of metal and run Linux.
Re:What are the Differences?
on
Free Solaris 8
·
· Score: 5
The most important difference is that Linux sucks substantially less.
More specific differences:
Linux runs on a proper superset of the platforms Solaris runs on.
Solaris is pretty strictly system V, while Linux is some SysV, some BSD, and some "other."
Solaris uses the UFS filesystem. Linux uses primarily the Ext2 filesystem
The obvious differences, like licensing (Linux is Free, Solaris is only pseudofree), sourcing (Solaris is single-source, Linux gives you choices), etc.
On common hardware, Linux tends to be faster, especially for interactive tasks.
(Personal observation) Solaris is stuck in the 70s and has obsolete administration tools; Linux is much more modern.
Solaris might be faster on machines with 16 processors or more.
If you need to do Java work, no surprise, the environment for Solaris is MUCH better.
Why is it people always say "Solaris x86 is slow cause, uh, the architecture sucks."?
Fact: I didn't say anything about slowness on x86, just that the x86 platform sucks. Which everyone knows. No Open Firmware, or anything close. 16-bit init code. IDE. Need I go on?
Furthermore, why do you call Solaris slow?
Because it is?
It scales very well.
Check out the HOARD project and see just how well it scales. And that's just multi-thread memory allocation. So it runs on 64-processor machines. So does UltraLinux. Big deal.
The file system is very fast. Maybe you find CDE slow?
The filesystem is fair-to-middlin at best. UFS is a basic Unix filesystem that could easily have come out of a university textbook. It can't touch xfs for streaming media applications and it can't touch ext2 for overall performance. Sure, it's faster than FAT, but that says nothing.
Actually, I don't find CDE at all. I refuse to pollute any system I'm responsible for with it. Thanks for nothing, HP.
That's funny. I've taken a fairly stock IRIX installation - without any other licenses or anything - and installed the gcc package from freeware.sgi.com on it. Then, I built my own gcc with it. Everything works smashingly. Judging by your post, I'd guess you left the old CAPS LOCK key on and were surprised to find that GCC, AS, and LD aren't on your system. SGI makes great stuff overall; I guess you just can't appreciate it. Enjoy your overclocked linux-powered 800 MHz gamez box, d00d.
Ignoring for a moment the question of whether the "buyers" have ever even seem $500k, you've made a grievous error:
If companies are going to be spending a half a million dollars on linux, and end up getting a product like "LinuxOne OS" which, according to this review doesn't work at all, they will be very unhappy with "linux" not with linux one.
So FUCKING what?
Really, will this affect your life one bit? NO! You will be every bit as free to use Linux as you've ever been. You will still find great support channels. You will still have access to all the source. You will still be free to browbeat your friends, neighbors, and children into using it. In short, your rights and freedoms will not be affected in the slightest.
Advocacy is one thing. Trying to restrict someone else's right to, within the confines of the licenses, distribute software in a way that is less than effective, is fundamentally anti-freedom. Your personal fear that "the product I advocate might get a bad name" is not justification for restricting someone else's freedom.
So don't buy stuff from these scumbags. Tell others to do the same. But don't whine about the harm they do as if it's the end of the world. It just isn't.
Enough LinuxOne stories already. We hate them. Fine. Everything that can be said already has been. Stop beating the pulp that was once a horse.
Sure, I'd just love to move from buggy-and-hard-to-install to slow-as-shit. I've used Solaris enough to say, in all honesty, that I think a company that made JUST Solaris would be worth even less than LinuxOne.
"Solaris - making your million-dollar hardware feel like a PDP-8 over a 75-baud line!"
"Solaris - forget asteroids through pipettes; we've redefined 'suck' altogether!"
"Solaris - proving again that bad designs refuse to die"
"Solaris - we'd love to be IRIX but we can't; we'd love to be linux but we won't"
Come on, Solaris people are worse than Linux people sometimes. All operating systems suck. Yours sucks a hell of a lot more than it needs to. Solaris on x86 suffers from the shittiness of the platform; Solaris on SPARC can't make that excuse. Solaris on SPARC is a criminal act punishable by death. If you want a machine that slow, I'll gladly trade you a shitty peecee w/linux for your beautiful Ultra {^5,10}.
Slashdot doesn't have an anti-Solaris conspiracy. Sun does. Get over it.
We're blaming these people for not actively seeking a better alternative?
Simply, yes. People who fail to do so make the world more hazardous for all of us by encouraging government regulation, monopolistic business practices, and other abominations. If we've got to live in a consumer-oriented economy, the least we can do is insist that consumers be smart about things. And failing to seek out the best is anything but. If you aren't going to research it, don't buy it. I fail to see the unfairness here.
We can all debate the evilness or non-evilness of AOL, but what's the point? Bottom line: the people having these problems are using an inferior operating system that allows these problems to occur by refusing to implement proper access control.
Looking at it another way, if I chmod -R 777 / and allow users to add and remove arbitrary kernel modules, and some user fux0rs my system, who's fault is that? Simple: it's my fault. Sure, properly behaved users and/or applications wouldn't hose it, but not everything is properly behaved. Some OS designers recognize that, and others don't. And the ones who don't just invite the kind of abuses AOL is perpetrating.
So don't blame AOL. They're just taking advantage of others' mistakes (both the OS designers' mistakes, and the users' mistakes for refusing to accept the reality - that the world is full of apps that can and will do Bad Things). So AOL is a trojan. Big deal. It's not the first trojan, and it won't be the last.
Is AOL evil? Probably. But the people who use it, and who use operating systems without access control, get exactly what they deserve.
-- TM, wondering why people care about this nonsense when they don't even use it
Nothing wrong with a negative post, but your complaint about lack of hard numbers is unjustified. Read the two benchamrking white papers, one on method and the other on specific results against a Mobile P3 @ 500. Not bad at all, really. If they can deliver, I don't think performance will be an issue.
The reason is that there is some rare hardware out there that corrupts data during IDE transfers when you either use DMA or receive an interrupt during a transfer.
And, ironically enough, most of the time this is caused by WD disks. Check the linux blacklist.
You know, it all sounds so good, until I remember that this is, in fact, reality and not fantasyland.
Try this: take a 6-year-old machine. (What kind? Your choice.) Put in an $80 scsi controller. Hook up some equally old monster diff scsi disks. A bit of software RAID, a few dozen knfsd's, and, bobsyouruncle, you're remote-installing operating systems on 40 machines at once. Smooth as silk and nice and fast. Saturates the internal disk bandwidth, the scsi bus, and the 100 mbit network, all at the same time as it should. Machines install, everyone goes home happy.
Now try this: Same thing, use ide controller(s) and disk(s). Watch while machine thrashes painfully and install-clients wait patiently (or not so) for their data. Network is mostly unused, bus sits idle. Not a balanced system at all.
It's easy to say IDE is fine when you've never really stressed anything. I know; I used to say that too. But it's just not true.
-- TM, wondering just how many O2's can install from a sparc 20 at once before Bad Things happen...
Finally, someone bothers to post some data. Fairly meaningless data, but at least a modicum of sanity. Thanks!
That said, I would question whether having extra ide channels is really any less expensive than scsi - especially when you consider that scsi also has standard external connectors and can expand to (in modern incarnations) 15 or more devices per channel vs 2 per channel, or only 1 if you want performance.
And before you all say, "It's part of the chipset, so it's basically free!" REMEMBER - Adding ide most definitely has a cost - for one, they could put scsi in instead. Of course, this would increase the complexity of the i/o chipset - almost to the point where the i/o processor is as important as the main cpu. Hmmm...what a concept.
3%? 25%? Of WHAT? An R2000? A P3? An ultrasparc? Nice "facts" in such an "insightful" post. Only trouble is, they're completely meaningless. You neglect to say what system this refers to, what configuration, what kind of adapters, what software and os; I could go on and on but why? This post is useless drivel. Please, don't post numbers if you haven't done benchmarks. "SCSI generally uses less main CPU time to do the same job" is correct. Your post is rubbish. See the difference?
2.The multitasking restriction is that IDE cannot issue more than 1 io request at a time, but this doesn't matter for single disk systems.
And in general, single disk systems are peecees, not workstations or servers. So, thanks for playing.
Yes, in many cases the drives are physically identical. So why don't we have 10k rpm ide drives? It might be marketing - or it might be that the vendors aren't going to waste the cost and effort to build those fast drives on ide. After all, systems with only ide are unlikely to get any increased benefit from additional media speed, and people who buy them aren't likely to be willing to pay the difference in disk cost.
You're forgetting the fundamental basis of peecee buyers: the only thing that matters is the ratio of $IMPORTANT_NUMBER to price. In this case, disk size. Nobody quotes MB/s or seek times or the crucial "platter to ethernet" time. Why? Because people buying biddy boxes don't give a fsck.
Slashdot Mirror
it runs on crappy, unscalable hardware.
If you have any of this, since it's so crappy, I'll happily take it off your hands. I'll even give you an enterprise-class (by YOUR definition) peecee for it.
it is not very efficient with its SMP code
Nothing is, sadly. It's a hard problem.
lets compare CDE to GNOME
Ok, lets. CDE: sucks 1e56 Ll. GNOME: sucks 1e56 Ll. Hmmm...I think I'll take "none of the above," please.
if you want a real enterprise os, buy redhat.
Define "enterprise." Red Hat is great for...well, it's ok for...ok, well, Red Hat is a big company. They stamp a lot of CDs. And the red hats are kind of cool. What were we talking about again?
Linux was modeled after Unix why again?? And Sun being the largest Unix distributor means??
Not a damn thing. Witness Microsoft. Lots o' market share, no product.
"Better to be thought an idiot that open your mouth and remove all doubt."
How apropos.
I know this isn't a support forum, but I do know such hardware pretty well. A few ideas: it may be that the machine is set to talk on serial ports instead of the GX. Try hooking up a terminal, 9600,8n1, and disconnecting the keyboard. If it works ok, see what input-device and output-device are set to ("printenv" at the prom prompt). If it's ttyX, you've found the problem. Do "setenv input-device keyboard" and "setenv output-device screen", plug the keyboard back in, reboot, and voila. Instant workstation. Or, try swapping out the GX for something else - even if you just have to borrow it. If the GX is truly fried, there's nothing you'll be able to do about it of course, so there's really nothing to do once you've found the source of the problem.
That's odd, you know; what I think of when comparing the two is Solaris and Linux on an identical midrange SPARC machine. An Ultra 2, 60, or 80, or a 450-ish machine. Of course, a comparison between the two on an identical peecee comes out about the same - Solaris loses. I don't see Solaris x86's shortcomings as being due to non-native hardware, but rather to inferior hardware, and Linux, to some degree, suffers from the same on that platform.
But on a real system, designed and crafted by men and women who know and love their trade, a system like one of the above, the peecee hardware barriers evaporate and the gloves come off. And the brilliant Sun engineers who designed and built those machines find out that their OS counterparts let them down. Badly. Have you ever run Linux on a SPARC? Even a low-end older system will work smashingly. But don't take this to mean that Linux can't take advantage of newer, faster hardware - it most certainly can. Unless you're one of a handful who need an E10k with dozens of processors, Linux is the only way to go on SPARCs. Try it; you'll never go back.
Don't take this the wrong way - Linux has its own problems. But those problems get fixed faster, and with better overall results, than the problems in Solaris. Is Linux the One and Only Solution? Not yet. But I can't find one instance where Solaris is really the best of breed. At the low end, lightweight products like QNX rule. In the low-to-midrange, Linux is king. And at the high end, IBM and SGI duke it out for the best massively scalable feature set. If you amalgamated the very best of all these into one product and called it Linux (for lack of a better name), I doubt you'd find any Solaris in it at all. But that's what you'll get, 10 years from now. And Solaris? A distant memory.
And of course, every year Linux squeezes more away from traditional vendors, who are left gasping for air at the summit of the high end, wondering whether this will be the year they finally have to throw in the towel and admit that they no longer make a product that people aren't running Linux on. Some, like SGI and IBM, have long since seen this coming, and realized that an OS is an expensive and unprofitable product to make, and taken steps to ensure the survival of their real business, hardware. At the same time they get to make sure that the best pieces of their system are incorporated into Linux so that their customers can continue to enjoy the very best. Maybe Sun just realizes they haven't anything to offer to the effort. After all, haven't they just given up on making money from Solaris? And yet they still refuse to unite behind the one product everyone else is agreeing on. There's a name for this: overweening pride.
Well, all I can say is that I'm going to miss Sun's hardware. When they finally follow the fading glory that was once SunOS into the great bitbucket in the sky, I can only hope the hardware teams land on their feet.
SunOS was BSD-based. Solaris is nearly pure System V. Have you ever actually used a BSD system?
My personal opinion of Solaris man pages is that they suck ass. Way, way too much space taken by tables with "SUNW,ZOIurgewqr" in them. Of course, since Solaris is seriously feature-starved, the man pages don't really have much to document, so they've got to fill them with something. But, again, that's just me. Now IRIX, there are some decent man pages.
To be fair, an E10k is an awfully nice machine. Linux would run on that too, if anybody could afford one. Your customer (and you!) can just be thankful that their application is suitable for clustering. Otherwise, the iron is essential. Operating system has little to do with it.
The only things Sun sells that anyone would want are made of metal and run Linux.
More specific differences:
Fact: I didn't say anything about slowness on x86, just that the x86 platform sucks. Which everyone knows. No Open Firmware, or anything close. 16-bit init code. IDE. Need I go on?
Furthermore, why do you call Solaris slow?
Because it is?
It scales very well.
Check out the HOARD project and see just how well it scales. And that's just multi-thread memory allocation. So it runs on 64-processor machines. So does UltraLinux. Big deal.
The file system is very fast. Maybe you find CDE slow?
The filesystem is fair-to-middlin at best. UFS is a basic Unix filesystem that could easily have come out of a university textbook. It can't touch xfs for streaming media applications and it can't touch ext2 for overall performance. Sure, it's faster than FAT, but that says nothing.
Actually, I don't find CDE at all. I refuse to pollute any system I'm responsible for with it. Thanks for nothing, HP.
That's funny. I've taken a fairly stock IRIX installation - without any other licenses or anything - and installed the gcc package from freeware.sgi.com on it. Then, I built my own gcc with it. Everything works smashingly. Judging by your post, I'd guess you left the old CAPS LOCK key on and were surprised to find that GCC, AS, and LD aren't on your system. SGI makes great stuff overall; I guess you just can't appreciate it. Enjoy your overclocked linux-powered 800 MHz gamez box, d00d.
If companies are going to be spending a half a million dollars on linux, and end up getting a product like "LinuxOne OS" which, according to this review doesn't work at all, they will be very unhappy with "linux" not with linux one.
So FUCKING what?
Really, will this affect your life one bit? NO! You will be every bit as free to use Linux as you've ever been. You will still find great support channels. You will still have access to all the source. You will still be free to browbeat your friends, neighbors, and children into using it. In short, your rights and freedoms will not be affected in the slightest.
Advocacy is one thing. Trying to restrict someone else's right to, within the confines of the licenses, distribute software in a way that is less than effective, is fundamentally anti-freedom. Your personal fear that "the product I advocate might get a bad name" is not justification for restricting someone else's freedom.
So don't buy stuff from these scumbags. Tell others to do the same. But don't whine about the harm they do as if it's the end of the world. It just isn't.
Enough LinuxOne stories already. We hate them. Fine. Everything that can be said already has been. Stop beating the pulp that was once a horse.
"Solaris - making your million-dollar hardware feel like a PDP-8 over a 75-baud line!"
"Solaris - forget asteroids through pipettes; we've redefined 'suck' altogether!"
"Solaris - proving again that bad designs refuse to die"
"Solaris - we'd love to be IRIX but we can't; we'd love to be linux but we won't"
Come on, Solaris people are worse than Linux people sometimes. All operating systems suck. Yours sucks a hell of a lot more than it needs to. Solaris on x86 suffers from the shittiness of the platform; Solaris on SPARC can't make that excuse. Solaris on SPARC is a criminal act punishable by death. If you want a machine that slow, I'll gladly trade you a shitty peecee w/linux for your beautiful Ultra {^5,10}.
Slashdot doesn't have an anti-Solaris conspiracy. Sun does. Get over it.
-- TM, who finds that Unix allows you to dodge the sticks, most of the time.
Simply, yes. People who fail to do so make the world more hazardous for all of us by encouraging government regulation, monopolistic business practices, and other abominations. If we've got to live in a consumer-oriented economy, the least we can do is insist that consumers be smart about things. And failing to seek out the best is anything but. If you aren't going to research it, don't buy it. I fail to see the unfairness here.
Looking at it another way, if I chmod -R 777 / and allow users to add and remove arbitrary kernel modules, and some user fux0rs my system, who's fault is that? Simple: it's my fault. Sure, properly behaved users and/or applications wouldn't hose it, but not everything is properly behaved. Some OS designers recognize that, and others don't. And the ones who don't just invite the kind of abuses AOL is perpetrating.
So don't blame AOL. They're just taking advantage of others' mistakes (both the OS designers' mistakes, and the users' mistakes for refusing to accept the reality - that the world is full of apps that can and will do Bad Things). So AOL is a trojan. Big deal. It's not the first trojan, and it won't be the last.
Is AOL evil? Probably. But the people who use it, and who use operating systems without access control, get exactly what they deserve.
-- TM, wondering why people care about this nonsense when they don't even use it
STOP! Pay troll.
Nothing wrong with a negative post, but your complaint about lack of hard numbers is unjustified. Read the two benchamrking white papers, one on method and the other on specific results against a Mobile P3 @ 500. Not bad at all, really. If they can deliver, I don't think performance will be an issue.
And, ironically enough, most of the time this is caused by WD disks. Check the linux blacklist.
By definition, no.
Try this: take a 6-year-old machine. (What kind? Your choice.) Put in an $80 scsi controller. Hook up some equally old monster diff scsi disks. A bit of software RAID, a few dozen knfsd's, and, bobsyouruncle, you're remote-installing operating systems on 40 machines at once. Smooth as silk and nice and fast. Saturates the internal disk bandwidth, the scsi bus, and the 100 mbit network, all at the same time as it should. Machines install, everyone goes home happy.
Now try this: Same thing, use ide controller(s) and disk(s). Watch while machine thrashes painfully and install-clients wait patiently (or not so) for their data. Network is mostly unused, bus sits idle. Not a balanced system at all.
It's easy to say IDE is fine when you've never really stressed anything. I know; I used to say that too. But it's just not true.
-- TM, wondering just how many O2's can install from a sparc 20 at once before Bad Things happen...
Of course there's a reason for higher prices - people are willing to pay them. What more reason do you need?
That said, I would question whether having extra ide channels is really any less expensive than scsi - especially when you consider that scsi also has standard external connectors and can expand to (in modern incarnations) 15 or more devices per channel vs 2 per channel, or only 1 if you want performance.
And before you all say, "It's part of the chipset, so it's basically free!" REMEMBER - Adding ide most definitely has a cost - for one, they could put scsi in instead. Of course, this would increase the complexity of the i/o chipset - almost to the point where the i/o processor is as important as the main cpu. Hmmm...what a concept.
3%? 25%? Of WHAT? An R2000? A P3? An ultrasparc? Nice "facts" in such an "insightful" post. Only trouble is, they're completely meaningless. You neglect to say what system this refers to, what configuration, what kind of adapters, what software and os; I could go on and on but why? This post is useless drivel. Please, don't post numbers if you haven't done benchmarks. "SCSI generally uses less main CPU time to do the same job" is correct. Your post is rubbish. See the difference?
And in general, single disk systems are peecees, not workstations or servers. So, thanks for playing.
Yes, in many cases the drives are physically identical. So why don't we have 10k rpm ide drives? It might be marketing - or it might be that the vendors aren't going to waste the cost and effort to build those fast drives on ide. After all, systems with only ide are unlikely to get any increased benefit from additional media speed, and people who buy them aren't likely to be willing to pay the difference in disk cost.
You're forgetting the fundamental basis of peecee buyers: the only thing that matters is the ratio of $IMPORTANT_NUMBER to price. In this case, disk size. Nobody quotes MB/s or seek times or the crucial "platter to ethernet" time. Why? Because people buying biddy boxes don't give a fsck.