It's been a while since I read much about IE7, but last I heard they were stripping a lot of its hooks out of the OS so that it sits "on top" like other browsers do.
What are these "hooks" you speak of ? What do you mean by "OS" ?
IE is - and always was - userspace code just like any other browser.
Oh, I'm not saying it's a bad interview; it's quite good. It just goes in a different direction than I think a slashdot interview would.
Well, yeah, but that's because a Slashdot interview would focus primarily on a software engineering decision made a decade ago and whether or not IE7 will support PNG transparency...
I didn't realize Gnome and KDE were Operating Systems.
I didn't realise that was relevant.
But then again, the difference between the GUI and the OS is a minor one, don't you think?
In the context of this discussion, it's unimportant. We are, after all, talking about user space shared components. How they are packaged and distributed to the end user simply doesn't matter, when you're looking at architecture and design.
Anybody using a UNIX system back then would know that it comes configured with rsh, telnet, rlogin, ftp, etc.
As would anyone using Windows NT4 and Windows 2000.
You need a password.
As you did with Windows NT4 and Windows 2000.
(note: the filesystem root is NOT exported by default)
You could typically *telnet as root* into unix machines that old by default. More recent ones you can typically "only" SSH in as root.
Now, which do you consider to be more exposure - the ability get a root shell or the ability to access a drive ?
Windows is a desktop and gaming OS.
Windows NT4 and 2000 were neither (although they could be used as such). They were business desktops targeted at managed environments. They were *never* targeted at home users and especially not for gaming.
It doesn't normally provide UNIX-style services.
No, it doesn't. However, Windows - even "home-user" Windows 3.x - had been providing filesharing out of the box since ca. 1993. You are talking about ca. 1997. It was a standard feature. Much like telnet, ftp, etc were on unix boxes.
I can't telnet in. Of course I don't expect to need a password.
You had to manually log in to Windows NT4. Windows 2000 did introduce an auto-login feature, but both required the setting of a password during installation. Windows NT4 was a multiuser, business OS designed to both use and provide network services. How could you *not* expect to "need a password" !?
Can you recommend some reading? I have a myriad of books on NT, 2000, XP and Windows 2003 Server (yes I do admin these systems) but even after reading extensive documentation, I still very much feel like there is a black box level that is inhibiting me from truly groking the system. Do realize that I have used Windows for far longer (~16 years) than *nix (about 6.5 years).
Not really. If you've trundled through numerous publications like "Inside Windows NT", along with MSDN and websites like sysinternals.com, there's not really much left short of getting a job at Microsoft;).
Windows is always going to fundamentally be a black box - I don't dispute that. What I do argue is that a) the average unix system is just as much a black box to the average unix administrator, even the more old-skool ones; and b) this doesn't have a meaningfully negative impact in the real world.
IMHO, adding users by running 'adduser blah' or even by manually editing/etc/passwd and friends is not meaningfully less abstracted than clicking "Add User" in Windows.
Obviously a difference of opinion. Having the option to dig down, debug and fix issues in the field is very powerful. While I don't have to do it often, when the need arises, it is really great to truly understand why an issue happens versus largely guessing, reinstalling or otherwise waiting for a support person to attempt to recreate the scenario and develop a solution. Sure, not all administrators have the skill set to do these things, but its nice not to be limited when the need arises.
I don't disagree it's powerful, I just think if it's something that is relied on, it is a flaw in the platform and leads to more problems than benefits.
I *should* be able to treat the machine as a black box, as should my users. Outside of exceptionally extraordinary circumstances (which as a non-specialised professional, I would not feel ashamed at being unable to address), intricate knowledge of system internals should not be necessary.
I can program competently, and have in the past. However, I certainly don't have the natural coding talent to grok the complex, non-trivial pieces of software in any sort of reasonable timeframe. Nor do I have the time, or - quite frankly - the inclination to spend doing so to fix something that shouldn't have broken in the first place.
Access to the source code is certainly a nice bonus, but I seriously question its usefulness for the vast majority of cases.
Basically, my position boils down to this: the more time I have to spend worrying about, learning about and dealing with, intricacies and technical details of a platform, the less time I have to be *using* that platform to provide faster, more reliable and more featureful services to my end users. Computers aren't hobbiest toys any more. It is no longer acceptable for the high priests of technology to have the only keys that unlock the productivity gains and lifestyle improvements computers offer. Modern platforms are simply too big and complex to expect more than a handful of people *anywhere* to be able to fully understand every aspect of them, and people shouldn't have to just so they can use them (and, yes, us sysadmins are users as well).
I live for the day when systems administrators will be anachronistic relics of the past - when computer errors only happen when the hardware is broken. Realisticall, I doubt that day will ever come, but I consider it the goal we should be striving for.
Indeed. I'm guessing these were fairly critical servers to be having a cluster/HA config. Thats one of the features of Linux.. you can get into that bleeding edge software (CVS code patches?!?) and end up finding yourself part of the development cycle. I really don't think thats a recommend approach for managing production level servers -- especially without testing on a dev server first.
That when you objectively listen to what *real people* from Microsoft are actually saying (and look at what they're doing), rather than apply biased feelings to out of context soundbites and "media analysts" with chips on their shoulders, they're just a bunch of geeks writing the best software they can.
I wonder what the average anti-Microsoftie would do if emails and memos from the "top brass" of Apple talked about how new features like Quartz, Expose, Spotlight and Boot Camp were going to "kill" Microsoft...
Argh. Must proofread. That should say:
I wonder what the average anti-Microsoftie would do if emails and memos "leaked" from the "top brass" of Apple talking about how new features like Quartz, Expose, Spotlight and Boot Camp were going to "kill" Microsoft...
Still in the process of watching it, but he has interesting perspective on Windows 95 and it's role as a bridge from 16bit to 32bit programs.
? Seems to me he was pretty much repeating what everyone (who was actually interested) already knew.
In fact, it might even help to get across to some people what a tremendous achievement Windows 95 actually was.
He also points out that though it wasnt the best OS they knew how to make at the time (points at NT) it was the best release of Windows that Microsoft ever did (in his opinion).
He said "successful", not "best". In fact, he specifically goes on to say that being "the best" in terms of technical features and design was very much a secondary priority.
OK, that should only be me. This was a dorm network, back when everyone else was running Windows 95. I certainly wasn't in any sort of realm or NT domain.
Computers aren't psychic. They have no way of knowing whether "you" really is "you" or someone pretending to be "you". All they can determine is whether or not some attempt to access resources has the right credentials.
Having not installed anything like telnet/rsh/ftp services, I shouldn't need a password.
Default configuration included file sharing. This is, I might point out, not markedly different from the default configuration of just about every similar OS from that era, many of which (particularly on the unix side) enabled far more "dangerous" remote services than filesharing.
You were using a particular piece of software outside of the default environment it was assumed it would be in. The onus was on *you* to reconfigure it appropriately. Again the computer (nor its developer) isn't psychic. It doesn't know what type of environment it's in.
This "security hole" is an order of magnitude less "serious" than allowing remote shell logins as root - something a non-trivial number of unixes do, even today. Don't try to blow it out of proportion just because you happened to get "cracked" due to operator error.
To me, he seems like a perfect example of a really smart person who doesn't understand that software is judged by how much easier it makes the user's life, not by how impressive the work is to his geek friends.
So you're saying he moonlights as an OSS developer ?
That's why internal memos and emails from the top brass (Ballmer and Gates, for instance) bragged to each other about how the IE integration was going to kill Netscape. Not because they wanted to kill the competition, but because they wanted to kill Netscape.
I wonder what the average anti-Microsoftie would do if emails and memos from the "top brass" of Apple talked about how new features like Quartz, Expose, Spotlight and Boot Camp were going to "kill" Microsoft...
The ideas of IE integration being both engineering *and* strategic business decision are not mutually exclusive.
The people responsible for the whole IE debacle (he actually uses this as an example) didn't integrate IE that way because they wanted to destroy the competition - they made an engineering decision at the time that they thought made sense and ended up causing a big brouhaha.
Not to mention an engineering decision that "everyone" (GNOME, KDE, OS X) subsequently copied. Doesn't that fit the category of "innovation" ?
The link you provided is quite disturbing. I feel like I must be reading it wrongly. That just can't be real. A plain old NT4 install would share all drives to the world without passwords?
You're reading it wrong. It will share it to Administrator level users (ie: you need an Administrator user/pass), for administrative purposes.
Probably the C$ administrative share had something to do with it. (WTF was that for anyway? I never asked for that. I disabled it many times, but Windows would helpfully restore the damn thing.)
I don't think he has ever really had an original design concept let alone created any product from scratch himself. ALL of Microsoft's products can be traced back to some other company.
Which company's products are you thinking of, that cannot ?
They saved themselves a lot of time by borrowing a lot from unix and nextOS, and reproduced some of the aspects of OS9.
They didn't "borrow it", they *bought it* [0]. OS X is NeXT with an updated kernel, warmed-over API, new display system and GUI. Basically, Apple did as much to NeXT to get OS X as Microsoft have done to Windows 2003 to get Vista. Which is to say, it was a pretty major update, but it wasn't a from-scratch, new OS.
Apple succesfully did what MS needs to do. They made a clean break with the past, knowing that although it might cause some problems up front, in the long run, it'd be for the best.
Apple did what Microsoft had done 5+ years earlier with Windows NT. Microsoft are lining up for their *second* "next generation OS" project - Apple are only just finishing their first.
Of course, Microsoft have to deal with a much larger customer base, so they work on much longer timescales - Apple aren't well-known for long transitional periods and dedication to legacy support, thus they can enact change faster.
[0] Not that there's anything wrong with doing this. It just seems to be a commonly held misconception that Apple "started from scratch" with OS X.
Others really should learn from that lesson of how to handle retiring archaic architecture that they don't want to drag along.
Well, Apple certainly did - their method of transitioning from MacOS to OS X was pretty much the same as Microsoft's transition from DOS-based Windows 3.x and Windows 95 to Windows NT, only with fewer phases, about 1/3 the timeframe and the benefit of computers ten times as powerful.
You forget, Microsoft was doing their "next generation" platform shuffle 5+ years before Apple.
Apple don't "retire" architectures, they kill them. The transitional periods are brutally short and usually come with little warning.
Microsoft, OTOH, swing too far in the other direction. Their transitional periods are too long (~10 years to get to Windows NT as the mainstream platform) and have so much warning that developers forget they're coming.
Here's an alternative that's a lot more likely to me.
[...]
You have pretty much described the Windows NT migration. This is wrapping up in Vista. Microsoft has been executing your "plan" for over a decade (which is how long it takes when you have to cater to 90% of the market).
However, you seem to think there's something wrong with the core of Windows NT and it needs to be rewritten. Why ?
Many have argued that Linus needs to stablize the kernel driver ABI. But on the other hand, by not doing so and encouraging drivers to be open source and in the kernel source tree brings us a large amount of stability that Windows just cannot achieve.
It's worth pointing out that pretty much every remotely mainstream OS *except* Linux manages to work (and work well) with a stable kernel ABI. Including ones considered at least - if not more - stable than Linux, even by Linux zealots, like FreeBSD and Solaris.
I'm guessing that would be IDE controllers? I haven't seen a SCSI RAID controller that works on anything less than the physical drive as the smallest granular unit. While I haven't played with the newest IDE RAID controllers, the ones from a couple of years ago were all based on whole disks.
Actually I can recall seeing this option on way more SCSI than IDE/SATA RAID controllers. Most IDE/SATA controllers I've used (eg: 3ware) only let you specify drives to create and array from and don't then allow you to choose the size of that array. SCSI RAID controllers tend to be more advanced (for obvious reasons).
The SATA RAID cards in our PE750s definitely allows this, and they're Adaptec derivatives. I'm pretty sure the ROMB in our PE[12][78]50s also allow it, although it has been a very long time since I've even tried to do such a thing (the utility of it is questionable in 99% of cases, and there is also the typically negative performance impact I mentioned before). A couple of big IBM Quad-Xeon boxes we setup in one of our US locations also offered that functionality (I remember this because one of the other admins setup "OS" and "Data" logical drives, something I considered to be pretty pointless given the conditions).
I should also note that I concluded that all the low-end IDE controllers are a waste of money compared to software RAID available with any decent OS. Once you get into the realm of hardware IDE RAID controllers that start to perform better than software raid, you wind up in the same price range as SCSI.
There are many scenarios for where disk performance is not critical, but reliability (and transparency) is. Not to mention ease of setup.
With that said, the performance of pretty much all non-RAID SATA and IDE controllers - once you get to a worthwhile density of 4+ drives per controller - is typically awful. Added to that, they usually don't play well with multiple cards per machine, or heavy simultaneous access from multiple drives (ie: software RAID usage). Even if you're only planning to use software RAID, you really need to buy "high end" RAID controllers like 3ware if you want a decent 8-port+ SATA controller.
(As I recently discovered when I though I'd try a couple of Promise SX8 cards for our newest disk array. Unfortunately, they suck, so now we have to shell out an additional AU$1200ish for a pair of 3ware 9550SX-8s - and that's assuming we can convince the vendor to let us return the Promise cards and get a refund, AU$1800 otherwise. I really think there's a market out there for solid, fast, non-RAID 8-port+ SATA controllers - if only someone would pander to it.)
So if it's just a fast mass store, perhaps a manual disk dupe snapshot is a better solution than an IDE raid. That's the route I've gone for my large data storage, and it makes for easy backups - no tapes, just extra drives that I copy and pull out whenever the data changes (this particular data doesn't change often, it's archival, so once the disk is full, it's done.)
The problem with your scenario is that when a disk fails, the whole machine is useless until the disk can be replaced, the machine reinstalled and backups restored. It's also a much less efficient use of space (vs RAID5 or RAID6). This might not be a problem in your environment, but rest assured it is in most:).
Personally I can't understand how anyone would even consider storing any remotely important business data on disks without RAID - even if it was only software RAID.
Slashdot Mirror
What are these "hooks" you speak of ? What do you mean by "OS" ?
IE is - and always was - userspace code just like any other browser.
Well, yeah, but that's because a Slashdot interview would focus primarily on a software engineering decision made a decade ago and whether or not IE7 will support PNG transparency...
I didn't realise that was relevant.
But then again, the difference between the GUI and the OS is a minor one, don't you think?
In the context of this discussion, it's unimportant. We are, after all, talking about user space shared components. How they are packaged and distributed to the end user simply doesn't matter, when you're looking at architecture and design.
As would anyone using Windows NT4 and Windows 2000.
You need a password.
As you did with Windows NT4 and Windows 2000.
(note: the filesystem root is NOT exported by default)
You could typically *telnet as root* into unix machines that old by default. More recent ones you can typically "only" SSH in as root.
Now, which do you consider to be more exposure - the ability get a root shell or the ability to access a drive ?
Windows is a desktop and gaming OS.
Windows NT4 and 2000 were neither (although they could be used as such). They were business desktops targeted at managed environments. They were *never* targeted at home users and especially not for gaming.
It doesn't normally provide UNIX-style services.
No, it doesn't. However, Windows - even "home-user" Windows 3.x - had been providing filesharing out of the box since ca. 1993. You are talking about ca. 1997. It was a standard feature. Much like telnet, ftp, etc were on unix boxes.
I can't telnet in. Of course I don't expect to need a password.
You had to manually log in to Windows NT4. Windows 2000 did introduce an auto-login feature, but both required the setting of a password during installation. Windows NT4 was a multiuser, business OS designed to both use and provide network services. How could you *not* expect to "need a password" !?
(Sorry for the slow response).
Can you recommend some reading? I have a myriad of books on NT, 2000, XP and Windows 2003 Server (yes I do admin these systems) but even after reading extensive documentation, I still very much feel like there is a black box level that is inhibiting me from truly groking the system. Do realize that I have used Windows for far longer (~16 years) than *nix (about 6.5 years).
Not really. If you've trundled through numerous publications like "Inside Windows NT", along with MSDN and websites like sysinternals.com, there's not really much left short of getting a job at Microsoft ;).
Windows is always going to fundamentally be a black box - I don't dispute that. What I do argue is that a) the average unix system is just as much a black box to the average unix administrator, even the more old-skool ones; and b) this doesn't have a meaningfully negative impact in the real world.
IMHO, adding users by running 'adduser blah' or even by manually editing /etc/passwd and friends is not meaningfully less abstracted than clicking "Add User" in Windows.
Obviously a difference of opinion. Having the option to dig down, debug and fix issues in the field is very powerful. While I don't have to do it often, when the need arises, it is really great to truly understand why an issue happens versus largely guessing, reinstalling or otherwise waiting for a support person to attempt to recreate the scenario and develop a solution. Sure, not all administrators have the skill set to do these things, but its nice not to be limited when the need arises.
I don't disagree it's powerful, I just think if it's something that is relied on, it is a flaw in the platform and leads to more problems than benefits.
I *should* be able to treat the machine as a black box, as should my users. Outside of exceptionally extraordinary circumstances (which as a non-specialised professional, I would not feel ashamed at being unable to address), intricate knowledge of system internals should not be necessary.
I can program competently, and have in the past. However, I certainly don't have the natural coding talent to grok the complex, non-trivial pieces of software in any sort of reasonable timeframe. Nor do I have the time, or - quite frankly - the inclination to spend doing so to fix something that shouldn't have broken in the first place.
Access to the source code is certainly a nice bonus, but I seriously question its usefulness for the vast majority of cases.
Basically, my position boils down to this: the more time I have to spend worrying about, learning about and dealing with, intricacies and technical details of a platform, the less time I have to be *using* that platform to provide faster, more reliable and more featureful services to my end users. Computers aren't hobbiest toys any more. It is no longer acceptable for the high priests of technology to have the only keys that unlock the productivity gains and lifestyle improvements computers offer. Modern platforms are simply too big and complex to expect more than a handful of people *anywhere* to be able to fully understand every aspect of them, and people shouldn't have to just so they can use them (and, yes, us sysadmins are users as well).
I live for the day when systems administrators will be anachronistic relics of the past - when computer errors only happen when the hardware is broken. Realisticall, I doubt that day will ever come, but I consider it the goal we should be striving for.
Indeed. I'm guessing these were fairly critical servers to be having a cluster/HA config. Thats one of the features of Linux .. you can get into that bleeding edge software (CVS code patches?!?) and end up finding yourself part of the development cycle. I really don't think thats a recommend approach for managing production level servers -- especially without testing on a dev server first.
Well, these weren't exactly
That when you objectively listen to what *real people* from Microsoft are actually saying (and look at what they're doing), rather than apply biased feelings to out of context soundbites and "media analysts" with chips on their shoulders, they're just a bunch of geeks writing the best software they can.
Argh. Must proofread. That should say:
I wonder what the average anti-Microsoftie would do if emails and memos "leaked" from the "top brass" of Apple talking about how new features like Quartz, Expose, Spotlight and Boot Camp were going to "kill" Microsoft...
? Seems to me he was pretty much repeating what everyone (who was actually interested) already knew.
In fact, it might even help to get across to some people what a tremendous achievement Windows 95 actually was.
He also points out that though it wasnt the best OS they knew how to make at the time (points at NT) it was the best release of Windows that Microsoft ever did (in his opinion).
He said "successful", not "best". In fact, he specifically goes on to say that being "the best" in terms of technical features and design was very much a secondary priority.
I'd have to say it depends on the individual.
If you're prone to drinking the Kool-aid, you're usually not discriminatory about whose you start drinking.
Computers aren't psychic. They have no way of knowing whether "you" really is "you" or someone pretending to be "you". All they can determine is whether or not some attempt to access resources has the right credentials.
Having not installed anything like telnet/rsh/ftp services, I shouldn't need a password.
Default configuration included file sharing. This is, I might point out, not markedly different from the default configuration of just about every similar OS from that era, many of which (particularly on the unix side) enabled far more "dangerous" remote services than filesharing.
You were using a particular piece of software outside of the default environment it was assumed it would be in. The onus was on *you* to reconfigure it appropriately. Again the computer (nor its developer) isn't psychic. It doesn't know what type of environment it's in.
This "security hole" is an order of magnitude less "serious" than allowing remote shell logins as root - something a non-trivial number of unixes do, even today. Don't try to blow it out of proportion just because you happened to get "cracked" due to operator error.
So you're saying he moonlights as an OSS developer ?
(Sorry, cheap shot - but I couldn't resist.)
I wonder what the average anti-Microsoftie would do if emails and memos from the "top brass" of Apple talked about how new features like Quartz, Expose, Spotlight and Boot Camp were going to "kill" Microsoft...
The ideas of IE integration being both engineering *and* strategic business decision are not mutually exclusive.
Not to mention an engineering decision that "everyone" (GNOME, KDE, OS X) subsequently copied. Doesn't that fit the category of "innovation" ?
Maybe because it's COMPLETELY NORMAL BEHAVIOUR ?
Teens having sex with teens is screwed up to begin with.
Why ? Because some fucked-up religious puritans think so ?
The link you provided is quite disturbing. I feel like I must be reading it wrongly. That just can't be real. A plain old NT4 install would share all drives to the world without passwords?
You're reading it wrong. It will share it to Administrator level users (ie: you need an Administrator user/pass), for administrative purposes.
History.
You probably got pwned by a weak Administrator password.
I would have thought a red-light camera does nothing more than providece evidence of an offense. Can you not challenge a ticket ?
Might not be the best course of action if you're in the US, however.
Wouldn't you need to know the "angle" of the polarisation filter for that to be effective ?
Which company's products are you thinking of, that cannot ?
They didn't "borrow it", they *bought it* [0]. OS X is NeXT with an updated kernel, warmed-over API, new display system and GUI. Basically, Apple did as much to NeXT to get OS X as Microsoft have done to Windows 2003 to get Vista. Which is to say, it was a pretty major update, but it wasn't a from-scratch, new OS.
Apple succesfully did what MS needs to do. They made a clean break with the past, knowing that although it might cause some problems up front, in the long run, it'd be for the best.
Apple did what Microsoft had done 5+ years earlier with Windows NT. Microsoft are lining up for their *second* "next generation OS" project - Apple are only just finishing their first.
Of course, Microsoft have to deal with a much larger customer base, so they work on much longer timescales - Apple aren't well-known for long transitional periods and dedication to legacy support, thus they can enact change faster.
[0] Not that there's anything wrong with doing this. It just seems to be a commonly held misconception that Apple "started from scratch" with OS X.
Well, Apple certainly did - their method of transitioning from MacOS to OS X was pretty much the same as Microsoft's transition from DOS-based Windows 3.x and Windows 95 to Windows NT, only with fewer phases, about 1/3 the timeframe and the benefit of computers ten times as powerful.
You forget, Microsoft was doing their "next generation" platform shuffle 5+ years before Apple.
Apple don't "retire" architectures, they kill them. The transitional periods are brutally short and usually come with little warning.
Microsoft, OTOH, swing too far in the other direction. Their transitional periods are too long (~10 years to get to Windows NT as the mainstream platform) and have so much warning that developers forget they're coming.
[...]
You have pretty much described the Windows NT migration. This is wrapping up in Vista. Microsoft has been executing your "plan" for over a decade (which is how long it takes when you have to cater to 90% of the market).
However, you seem to think there's something wrong with the core of Windows NT and it needs to be rewritten. Why ?
It's worth pointing out that pretty much every remotely mainstream OS *except* Linux manages to work (and work well) with a stable kernel ABI. Including ones considered at least - if not more - stable than Linux, even by Linux zealots, like FreeBSD and Solaris.
Actually I can recall seeing this option on way more SCSI than IDE/SATA RAID controllers. Most IDE/SATA controllers I've used (eg: 3ware) only let you specify drives to create and array from and don't then allow you to choose the size of that array. SCSI RAID controllers tend to be more advanced (for obvious reasons).
The SATA RAID cards in our PE750s definitely allows this, and they're Adaptec derivatives. I'm pretty sure the ROMB in our PE[12][78]50s also allow it, although it has been a very long time since I've even tried to do such a thing (the utility of it is questionable in 99% of cases, and there is also the typically negative performance impact I mentioned before). A couple of big IBM Quad-Xeon boxes we setup in one of our US locations also offered that functionality (I remember this because one of the other admins setup "OS" and "Data" logical drives, something I considered to be pretty pointless given the conditions).
I should also note that I concluded that all the low-end IDE controllers are a waste of money compared to software RAID available with any decent OS. Once you get into the realm of hardware IDE RAID controllers that start to perform better than software raid, you wind up in the same price range as SCSI.
There are many scenarios for where disk performance is not critical, but reliability (and transparency) is. Not to mention ease of setup.
With that said, the performance of pretty much all non-RAID SATA and IDE controllers - once you get to a worthwhile density of 4+ drives per controller - is typically awful. Added to that, they usually don't play well with multiple cards per machine, or heavy simultaneous access from multiple drives (ie: software RAID usage). Even if you're only planning to use software RAID, you really need to buy "high end" RAID controllers like 3ware if you want a decent 8-port+ SATA controller.
(As I recently discovered when I though I'd try a couple of Promise SX8 cards for our newest disk array. Unfortunately, they suck, so now we have to shell out an additional AU$1200ish for a pair of 3ware 9550SX-8s - and that's assuming we can convince the vendor to let us return the Promise cards and get a refund, AU$1800 otherwise. I really think there's a market out there for solid, fast, non-RAID 8-port+ SATA controllers - if only someone would pander to it.)
So if it's just a fast mass store, perhaps a manual disk dupe snapshot is a better solution than an IDE raid. That's the route I've gone for my large data storage, and it makes for easy backups - no tapes, just extra drives that I copy and pull out whenever the data changes (this particular data doesn't change often, it's archival, so once the disk is full, it's done.)
The problem with your scenario is that when a disk fails, the whole machine is useless until the disk can be replaced, the machine reinstalled and backups restored. It's also a much less efficient use of space (vs RAID5 or RAID6). This might not be a problem in your environment, but rest assured it is in most :).
Personally I can't understand how anyone would even consider storing any remotely important business data on disks without RAID - even if it was only software RAID.