...but I haven't seen any kind of shortage. You can find the 360 at pretty much any decent retailer. These things aren't exactly jumping off the shelves here (Southern USA). Everyone that I know that was planning on getting one right away has gotten one, without standing in God-awfully long lines or any of the other stuff I have heard of people going through to get one. They bought them at normal retail prices, and didn't resort to paying $5000 on Ebay. Perhaps it is just the extremes (rural and urban) that don't have any of them?
Windows 2000/2003/XP have IPv6 support (though 2000's isn't the greatest) Linux/*BSD/MacOS have IPv6 support Cisco/Juniper/Foundry have IPv6 support
Not only have these operating systems and devices *had* IPv6 support, but most of them have had it for years already. Just because people don't turn it on doesn't mean the device cannot handle it. I wouldn't be surprised if we came up with a massive amount of IPv6 penetration already if devices just turned it on by default. Sure some of your cheaper devices don't support it, but most end-user type devices are field-upgradable for exactly this reason. Heck, my *cell phone* is capable of IPv6. What is really stopping IPv6 adoption is that there's no compelling reason to use it. You don't get access to new content. You don't get access to anything that you cannot reach via IPv4, under most cases.
I think you're missing the point that "private sector" does not equate to "start-ups" in any way. It could very easily be the same people that built the shuttle that undertake this. The difference would be that it's their dime when they launch the thing, so it would be much more efficient and probably more sturdy. After all, they would be designing it form the ground up to be used by themselves, so they would want to be able to maximize profits and make it as reusable and safe as possible. Not add on possible scenarios of use such as having it as a mobile weapons launching platform. Who exactly do you think builds all of these shuttles/space planes/etc for NASA? The private sector, of course.
Valve has decided to switch focus from the First-Person Shooter Genre, the only place they have had any success at all, and to spend time developing games that feature cooperative tasks that nobody does in real life and are boring as shit when made into a game. Valve is hoping to be the first company to release a MMOWP, a Massively Multiplayer Online Work Place. They hope to leverage the eWorkers into a viable force to artificailly explore the solar system and other stuff that their graphics guys get their jollies off on drawing. Sources say that Half-Life 3 will be a cooperative game where players count the vibration of radioactive materials and collectively calculate their half lives.
Ask yourself this as well: when was the last time an open-source project you help out with surveyed its users to find out what was most important to THEM?
In fact, it is very common for us to do this between all of our releases. We take these feature requests, along with the bug reports generated between the release, and try to work towards resolving every last one that we can before we release.
Ehh... no. See, that's what all the Gentoo-hating asshats that have never even used Gentoo tell you. Well, either them or the even-more-annoying ricers that think that by compiling their system with -ffast-math they're actually doing something productive.
So what is the advantage of installing Gentoo from binaries?
Well, you get a completely working system up and usable very quickly. All updates are then done from source. You can optimize your system on your own time table while still having a usable system. You also get the portage package manager, which is arguably the best out there.
Only complete idiots think that everything in Gentoo is about optimizations. The primary reason most people use Gentoo is for the customization ability. Want a lightweight version of a package and don't need to have optional ldap, mysql, and postgresql back-ends compiled into your mail client? No problem. To be honest, I would say that more load time is saved by having smaller binaries by stripping out optional support the user doesn't want, than is ever gained through aggressive optimizations. At any rate, most Gentoo users have enough sense not to overoptimize their binaries. It's just the vocal few that give the rest of us a bad name. Kind of like how a very small minority of Debian users have a seriously elitist attitude, so people relate Debian to being a bunch of elitist zealots.
It is the people that continue to propogate these myths that are the real bane to both communities.
Agreed 100%. I am also not a complete zealot, like some Linux users. I will gladly take half-binary/half-source drivers like NVidia's or ATI's. I will gladly take a binary-only driver, provided it is a userland driver. The only reason I wouldn't like a binary-only kernel driver is because it likely won't work with many kernels.
I just want my hardware to work and I want the company that I am buying it from to support me.
Of course, I would love for the company to release specs, or create their own open source drivers, but in the absence of these, I just want a driver that works.
If you're playing Windows games, then you aren't Windows-free in any aspect. You are still supporting Windows and Windows gaming. Using Cedega is a win for Microsoft, not for Linux.
Free software can be distributed and it's up to the distros to make.debs,.rpms,.ebuilds, etc. But how do you do that with something like photoshop or illustrator? People want to buy the CD and pop it into any computer and install it. And adobe can't possibly make a different package for each distro, even the popular ones.
Why not? You have to support exactly 1 scenario to cover most of the "popular" ones. Two, if you're feeling generous. RPM and DEB. That's it. It is not hard, at all, to create a RPM that works properly on Red Hat/SuSE/Mandriva.
I'm using gentoo and ubuntu right now. I love them because thousands of software titles are available either with the click of the mouse or a few keystrokes in a console. But this works because people get those free packages and configure them for each distro either because their distro paid them or out of the goodness of their heart. But it'd be illegal for someone to make a photoshop ebuild that distributed all the files. And it's a pain to copy the photoshop files into/usr/portage/distfiles and have an ebuild work from there (as in sun-jdk, crossover office, etc.).
You must mean like the multi-thousand dollar Maya that can be installed with the following commands on Gentoo:
Now, Maya is only distributed as an RPM on CD, and UT2004 is distributed as a Loki_Setup distribution wrapped with MakeSelf along with data on the DVD. The games that are ported to Linux don't seem to have a problem being supported, why would other pieces of commercial software be any different? Is it really that hard to have a "setup.sh" on the CD? I've never understood the complexity problem that people seem to think is there. Is running "setup.sh" and clicking "Next" a few times really that much harder than running "setup.exe" and clicking "Next" a few times? Why is it that Veritas/Oracle/Epic/Id and others can get this and Adobe cannot?
I couldn't agree more. When I switch from using my AMD64 machines at home to my work machine running a 32-bit Linux, I find myself wishing that I had no flash plugin on the work machine. Simply browsing the web is much slower and takes up more resources due to flash ads. I also find that I'm not missing out on much by not having flash. I mean, sure, I miss out on some of the dumb joke web pages that people send out to me via email, but I can always watch it on another computer. I've just found my browsing experience to be much more enjoyable without it.
2)Gentoo had the same bug in their ready-for-the-world live gaming CDs (and hadn't tracked it down).
No. Gentoo did not have any such bug, but only because Gentoo did not release those CD images. It was Daniel Robbins under another commercial venture, called Gentoo Games, where he was leveraging the Gentoo name to turn a profit for himself. These were not a part of the community distribution, and even the distribution developers were not told about it until the release date. We found out about it when it hit the front page of Gentoo's website just like everyone else. Also remember that Daniel, through this venture, violated Id Software's copyright on Enemy Territory by distributing it in a modified form, without permission of Id Software. Gentoo Games also provided no support for these discs, and tried to dump the support on the Gentoo Linux developers. Needless to say, Gentoo Games died a very quick and uneventful death.
I posted this yesterday
( " The New FreeBSD Website is UP Thursday October 06, @06:15AM Rejected" )
as news only to be rejected. I dont know why it was rejected so i cant complain i was treated unfairly. But when someone posts news and is rejected then the news appears a day later posted by someone else. It makes me wonder what the fsck is going on round here.
Well, with a multitude of editors and tons of people submitting articles, they simply just pick them at random. You might have been the first, you might have been the last. It doesn't matter. They simply pick one. This is also one of the main reasons for duplications. So many people post variations on the same thing, and with the multiple "editors" on the site, things simply get posted by more than one of them. The only real way I can see to improve this situation would be to get people that actually check the site for duplicates, and check the submission queue for duplicates, and pick the best submission for a topic. I think time constraints are probably the main reason this isn't done.
When Gentoo makes a new release, we submit a story to Slashdot. Since we know about our releases well ahead of any users, you would think that our submission would get used. It never has. We even go so far as to make sure our Slashdot submission is more of a teaser/summery, than a full-blown press release, as I could understand not wanting to post something that reads identical to the press release. Instead, hours and hours later, we usually get a posting that was submitted by a user, is chock full of false statements and half-truths, and doesn't point out anything that would actually be of interest to anyone.
What do we do about it?
Nothing. We understand that this is the nature of Slashdot, and we submit another story the next release.
I do think the new site looks awesome. Great job, FreeBSD and Emily!
Paying a developer to spend most of their time working for you on your proprietary back-end/astroturfing, lowering the volume of bugs that they resolve tenfold, is not "funded paying for Bugs in Gentoo for almost 8 months" at all.
Actually, the business is geared towards smaller outfits. Maybe not one without an admin, but perhaps one with only a single admin and older computers. In fact, at their presentation at the Gentoo Developer Conference after LWE:SF, they specifically mentioned older machines in their presentation. You can view their entire presentation online at http://devconference.gentoo.org/ (warning, streaming video). They were last in the afternoon session.
Please. Everyone knows that Hugh Hefner isn't a pornagrapher. He's just a very cool, very perverted old man with 7 hot blonde girlfriends who happens to run one of the most successful magazine publishers on the planet and has probably single-handedly helped America get to the state it is in today in regards to sexuality. With how repressed America is *even with* Hugh, I'd hate to see how we would have turned out without him. Besides, you have to admire a guy that lets us jerk off to pictures of girls he's already fucked. *grin*
What amazes me is that they had IBM hardware and RedHat Engineers working on this and it still didn't work. I've installed Linux servers for 10 years and rarely have experienced such problems. Usually it was the hardware or my screw up at the center of it all.
Where I work, we run IBM BladeCenters with HS20 blades. We've had *tons* of issues with Red Hat Enterprise Linux on these machines. Now, the issues have been worked out over the past couple years, but getting fixes for some of these items has taken months. Almost all of the issues have ended up being kernel-related, due to patches added to the kernel by Red Hat. The remaining issues were due to poor default configurations, such as having crap like kudzu enabled by default on a server OS. Honestly, I don't think Red Hat has the support staff to be able to properly support all of the plethora of hardware out there efficiently.
Well, quite. Which means that there are a whole class of problems that can't be addressed by third party specialists if you use a proprietory OS. The same for apps too, of course.
It's a lot less than you think. I've dealt with IBM before. I've even had IBM issue me Windows patches to resolve an issue. If you deal with a big enough fish, they probably have access to Microsoft's source. I'm no fan of Microsoft, but trying to make it out like their closed-source nature removes support options isn't exactly a factual statement. Sure, you might not be able to hire some kid you know from down the street to whip up a patch for your mission critical server, but what company in their right mind would do such a thing to begin with?
This has got to be one of the stupidest things I've ever heard. No offense, buddy, but almost nobody gets support from Microsoft. They usually get their support from third parties or manage it themselves with Windows. What exactly do you think those IT people *do* all day long? The primary reasons for not using Microsoft for support is the cost and the unresponsiveness. Ever been on the phone with a Microsoft support technician, who's on a laggy VoIP connection from Bangalore or wherever "Ed" claims to be from? Talking to their support is a nightmare. This is exactly why guys like IBM manage to make money hand over fist doing support. They're simply better at it.
Having the source doesn't change the fact that you're going to have to pay *somebody* to support it.
This is absolutely wrong. Politics are vital in most business. In fact, it is politics that keeps people from building sites that are friendly to non-IE browsers. Management wants something done fast more than they want something done right. Besides, corporations are so hell-bent on the short-term that they can't see far enough into the future to realize when they're screwing themselves. Who cares if we'll save $100,000 in labor costs over the next year if we can save $20,000 on next quarter's bottom line? Sadly, this is the short-sightedness prevalent in most companies. The worst thing is that the guy that makes this "brilliant" choice for the company will probably be promoted, and allowed to make even more "brilliant" choices in the future.
Most likely, your land line *is* VoIP, unless you're running through one of the slow-moving ILECs. I work for a telecommunications company and *all* calls that traverse our network end up being VoIP at some point. Our entire backbone is VoIP. Sure, you might have an analog signal to the CO, but after that, you're VoIP. We even offer VoIP services. That being said, we only cater to businesses, so we don't run VoIP over crappy broadband, but instead over SLA-backed services such as T-1 and T-3 lines. The major advantage to our VoIP services is it is regulated on the line by hardware, so if you aren't making any calls, you have the full bandwidth for downloads. QoS is ensured on the network, so voice calls always come first. I can say that with our system I have seen exactly zero loss of quality, and sometimes it even sounds better than classic analog lines. Remember that VoIP itself isn't the problem, it's crappy end-to-end IP connectivity causing the VoIP issues.
This is basically the priciple that my company runs on with their servers. We should be able to be running perfectly fine with 2/5ths of our servers down at any given time. Of course, this almost never happens, but building with that sort fo redundancy in mind reduces the chances of downtime to almost nothing. Each machine is also on redundant links to redundant switches on redundant upstream links. We do have the advantage of being an ISP and CLEC ourselves, so we already have multiple peering agreements with many other CLEC/ILECs.
As for the double-failure in a RAID5 array thing the article poster mentioned, for Pete's sake, buy a couple spare disks. You should follow the same rule in making your RAID arrays as your server clusters. You *should* be able to lose 2/5ths of your disks without losing the array. This means that you need at least 1 spare for every 5 drives, for a total of 6 drives.
Add some good monitoring on top of these and your downtimes drop to almost nothing. In fact, you shouldn't ever see service downtimes with a proper setup, provided you actually bring machines back up as they fail.
Slashdot Mirror
...but I haven't seen any kind of shortage. You can find the 360 at pretty much any decent retailer. These things aren't exactly jumping off the shelves here (Southern USA). Everyone that I know that was planning on getting one right away has gotten one, without standing in God-awfully long lines or any of the other stuff I have heard of people going through to get one. They bought them at normal retail prices, and didn't resort to paying $5000 on Ebay. Perhaps it is just the extremes (rural and urban) that don't have any of them?
Not at most places...
Windows 2000/2003/XP have IPv6 support (though 2000's isn't the greatest)
Linux/*BSD/MacOS have IPv6 support
Cisco/Juniper/Foundry have IPv6 support
Not only have these operating systems and devices *had* IPv6 support, but most of them have had it for years already. Just because people don't turn it on doesn't mean the device cannot handle it. I wouldn't be surprised if we came up with a massive amount of IPv6 penetration already if devices just turned it on by default. Sure some of your cheaper devices don't support it, but most end-user type devices are field-upgradable for exactly this reason. Heck, my *cell phone* is capable of IPv6. What is really stopping IPv6 adoption is that there's no compelling reason to use it. You don't get access to new content. You don't get access to anything that you cannot reach via IPv4, under most cases.
I think you're missing the point that "private sector" does not equate to "start-ups" in any way. It could very easily be the same people that built the shuttle that undertake this. The difference would be that it's their dime when they launch the thing, so it would be much more efficient and probably more sturdy. After all, they would be designing it form the ground up to be used by themselves, so they would want to be able to maximize profits and make it as reusable and safe as possible. Not add on possible scenarios of use such as having it as a mobile weapons launching platform. Who exactly do you think builds all of these shuttles/space planes/etc for NASA? The private sector, of course.
Dude... we're talking the *freakin Internet* man... they use switches, not hubs... ;P
Valve has decided to switch focus from the First-Person Shooter Genre, the only place they have had any success at all, and to spend time developing games that feature cooperative tasks that nobody does in real life and are boring as shit when made into a game. Valve is hoping to be the first company to release a MMOWP, a Massively Multiplayer Online Work Place. They hope to leverage the eWorkers into a viable force to artificailly explore the solar system and other stuff that their graphics guys get their jollies off on drawing. Sources say that Half-Life 3 will be a cooperative game where players count the vibration of radioactive materials and collectively calculate their half lives.
About 24 hours ago.
http://article.gmane.org/gmane.linux.gentoo.releng /375
I also did it about 36 weeks ago.
http://article.gmane.org/gmane.linux.gentoo.releng /174
In fact, it is very common for us to do this between all of our releases. We take these feature requests, along with the bug reports generated between the release, and try to work towards resolving every last one that we can before we release.
You're missing the whole point of this great addition to Skype... Cam Whores...
Why else does anyone need to see the person on the other end than to see them getting naked?
A trojan to redirect my browser to porn sites. I do that well enough without the assistance. *grin*
Ehh... no. See, that's what all the Gentoo-hating asshats that have never even used Gentoo tell you. Well, either them or the even-more-annoying ricers that think that by compiling their system with -ffast-math they're actually doing something productive.
So what is the advantage of installing Gentoo from binaries?
Well, you get a completely working system up and usable very quickly. All updates are then done from source. You can optimize your system on your own time table while still having a usable system. You also get the portage package manager, which is arguably the best out there.
Only complete idiots think that everything in Gentoo is about optimizations. The primary reason most people use Gentoo is for the customization ability. Want a lightweight version of a package and don't need to have optional ldap, mysql, and postgresql back-ends compiled into your mail client? No problem. To be honest, I would say that more load time is saved by having smaller binaries by stripping out optional support the user doesn't want, than is ever gained through aggressive optimizations. At any rate, most Gentoo users have enough sense not to overoptimize their binaries. It's just the vocal few that give the rest of us a bad name. Kind of like how a very small minority of Debian users have a seriously elitist attitude, so people relate Debian to being a bunch of elitist zealots.
It is the people that continue to propogate these myths that are the real bane to both communities.
Yup. I'm pretty sure that he is saying exactly that.
Agreed 100%. I am also not a complete zealot, like some Linux users. I will gladly take half-binary/half-source drivers like NVidia's or ATI's. I will gladly take a binary-only driver, provided it is a userland driver. The only reason I wouldn't like a binary-only kernel driver is because it likely won't work with many kernels.
I just want my hardware to work and I want the company that I am buying it from to support me.
Of course, I would love for the company to release specs, or create their own open source drivers, but in the absence of these, I just want a driver that works.
If you're playing Windows games, then you aren't Windows-free in any aspect. You are still supporting Windows and Windows gaming. Using Cedega is a win for Microsoft, not for Linux.
Why not? You have to support exactly 1 scenario to cover most of the "popular" ones. Two, if you're feeling generous. RPM and DEB. That's it. It is not hard, at all, to create a RPM that works properly on Red Hat/SuSE/Mandriva.
You must mean like the multi-thousand dollar Maya that can be installed with the following commands on Gentoo:
How about Unreal Tournament 2004? (DVD just to make it simpler)
Now, Maya is only distributed as an RPM on CD, and UT2004 is distributed as a Loki_Setup distribution wrapped with MakeSelf along with data on the DVD. The games that are ported to Linux don't seem to have a problem being supported, why would other pieces of commercial software be any different? Is it really that hard to have a "setup.sh" on the CD? I've never understood the complexity problem that people seem to think is there. Is running "setup.sh" and clicking "Next" a few times really that much harder than running "setup.exe" and clicking "Next" a few times? Why is it that Veritas/Oracle/Epic/Id and others can get this and Adobe cannot?
I couldn't agree more. When I switch from using my AMD64 machines at home to my work machine running a 32-bit Linux, I find myself wishing that I had no flash plugin on the work machine. Simply browsing the web is much slower and takes up more resources due to flash ads. I also find that I'm not missing out on much by not having flash. I mean, sure, I miss out on some of the dumb joke web pages that people send out to me via email, but I can always watch it on another computer. I've just found my browsing experience to be much more enjoyable without it.
No. Gentoo did not have any such bug, but only because Gentoo did not release those CD images. It was Daniel Robbins under another commercial venture, called Gentoo Games, where he was leveraging the Gentoo name to turn a profit for himself. These were not a part of the community distribution, and even the distribution developers were not told about it until the release date. We found out about it when it hit the front page of Gentoo's website just like everyone else. Also remember that Daniel, through this venture, violated Id Software's copyright on Enemy Territory by distributing it in a modified form, without permission of Id Software. Gentoo Games also provided no support for these discs, and tried to dump the support on the Gentoo Linux developers. Needless to say, Gentoo Games died a very quick and uneventful death.
Well, with a multitude of editors and tons of people submitting articles, they simply just pick them at random. You might have been the first, you might have been the last. It doesn't matter. They simply pick one. This is also one of the main reasons for duplications. So many people post variations on the same thing, and with the multiple "editors" on the site, things simply get posted by more than one of them. The only real way I can see to improve this situation would be to get people that actually check the site for duplicates, and check the submission queue for duplicates, and pick the best submission for a topic. I think time constraints are probably the main reason this isn't done.
When Gentoo makes a new release, we submit a story to Slashdot. Since we know about our releases well ahead of any users, you would think that our submission would get used. It never has. We even go so far as to make sure our Slashdot submission is more of a teaser/summery, than a full-blown press release, as I could understand not wanting to post something that reads identical to the press release. Instead, hours and hours later, we usually get a posting that was submitted by a user, is chock full of false statements and half-truths, and doesn't point out anything that would actually be of interest to anyone.
What do we do about it?
Nothing. We understand that this is the nature of Slashdot, and we submit another story the next release.
I do think the new site looks awesome. Great job, FreeBSD and Emily!
Paying a developer to spend most of their time working for you on your proprietary back-end/astroturfing, lowering the volume of bugs that they resolve tenfold, is not "funded paying for Bugs in Gentoo for almost 8 months" at all.
Actually, the business is geared towards smaller outfits. Maybe not one without an admin, but perhaps one with only a single admin and older computers. In fact, at their presentation at the Gentoo Developer Conference after LWE:SF, they specifically mentioned older machines in their presentation. You can view their entire presentation online at http://devconference.gentoo.org/ (warning, streaming video). They were last in the afternoon session.
Please. Everyone knows that Hugh Hefner isn't a pornagrapher. He's just a very cool, very perverted old man with 7 hot blonde girlfriends who happens to run one of the most successful magazine publishers on the planet and has probably single-handedly helped America get to the state it is in today in regards to sexuality. With how repressed America is *even with* Hugh, I'd hate to see how we would have turned out without him. Besides, you have to admire a guy that lets us jerk off to pictures of girls he's already fucked. *grin*
Hugh Hefner for president!
Where I work, we run IBM BladeCenters with HS20 blades. We've had *tons* of issues with Red Hat Enterprise Linux on these machines. Now, the issues have been worked out over the past couple years, but getting fixes for some of these items has taken months. Almost all of the issues have ended up being kernel-related, due to patches added to the kernel by Red Hat. The remaining issues were due to poor default configurations, such as having crap like kudzu enabled by default on a server OS. Honestly, I don't think Red Hat has the support staff to be able to properly support all of the plethora of hardware out there efficiently.
Well, quite. Which means that there are a whole class of problems that can't be addressed by third party specialists if you use a proprietory OS. The same for apps too, of course.
It's a lot less than you think. I've dealt with IBM before. I've even had IBM issue me Windows patches to resolve an issue. If you deal with a big enough fish, they probably have access to Microsoft's source. I'm no fan of Microsoft, but trying to make it out like their closed-source nature removes support options isn't exactly a factual statement. Sure, you might not be able to hire some kid you know from down the street to whip up a patch for your mission critical server, but what company in their right mind would do such a thing to begin with?
This has got to be one of the stupidest things I've ever heard. No offense, buddy, but almost nobody gets support from Microsoft. They usually get their support from third parties or manage it themselves with Windows. What exactly do you think those IT people *do* all day long? The primary reasons for not using Microsoft for support is the cost and the unresponsiveness. Ever been on the phone with a Microsoft support technician, who's on a laggy VoIP connection from Bangalore or wherever "Ed" claims to be from? Talking to their support is a nightmare. This is exactly why guys like IBM manage to make money hand over fist doing support. They're simply better at it.
Having the source doesn't change the fact that you're going to have to pay *somebody* to support it.
This is absolutely wrong. Politics are vital in most business. In fact, it is politics that keeps people from building sites that are friendly to non-IE browsers. Management wants something done fast more than they want something done right. Besides, corporations are so hell-bent on the short-term that they can't see far enough into the future to realize when they're screwing themselves. Who cares if we'll save $100,000 in labor costs over the next year if we can save $20,000 on next quarter's bottom line? Sadly, this is the short-sightedness prevalent in most companies. The worst thing is that the guy that makes this "brilliant" choice for the company will probably be promoted, and allowed to make even more "brilliant" choices in the future.
Most likely, your land line *is* VoIP, unless you're running through one of the slow-moving ILECs. I work for a telecommunications company and *all* calls that traverse our network end up being VoIP at some point. Our entire backbone is VoIP. Sure, you might have an analog signal to the CO, but after that, you're VoIP. We even offer VoIP services. That being said, we only cater to businesses, so we don't run VoIP over crappy broadband, but instead over SLA-backed services such as T-1 and T-3 lines. The major advantage to our VoIP services is it is regulated on the line by hardware, so if you aren't making any calls, you have the full bandwidth for downloads. QoS is ensured on the network, so voice calls always come first. I can say that with our system I have seen exactly zero loss of quality, and sometimes it even sounds better than classic analog lines. Remember that VoIP itself isn't the problem, it's crappy end-to-end IP connectivity causing the VoIP issues.
This is basically the priciple that my company runs on with their servers. We should be able to be running perfectly fine with 2/5ths of our servers down at any given time. Of course, this almost never happens, but building with that sort fo redundancy in mind reduces the chances of downtime to almost nothing. Each machine is also on redundant links to redundant switches on redundant upstream links. We do have the advantage of being an ISP and CLEC ourselves, so we already have multiple peering agreements with many other CLEC/ILECs.
As for the double-failure in a RAID5 array thing the article poster mentioned, for Pete's sake, buy a couple spare disks. You should follow the same rule in making your RAID arrays as your server clusters. You *should* be able to lose 2/5ths of your disks without losing the array. This means that you need at least 1 spare for every 5 drives, for a total of 6 drives.
Add some good monitoring on top of these and your downtimes drop to almost nothing. In fact, you shouldn't ever see service downtimes with a proper setup, provided you actually bring machines back up as they fail.