Doesn't this strike you as ironic? At first glance, sure. But when you realize that the bandwidth consumed is causing other research and educational use of the network to be severely degraded, it eases the irony.
If you want to test a new distribution scheme, you can use a lab setup. If you need a lot of nodes, you can use http://www.planet-lab.org/. If 99.9% of the use of the new technology has nothing to do with research, I'm not sure I see the value in letting that use hamper the rest of the research at your institution.
Except that the cost to OEMs is significantly cheaper. Ever get a refund for not using the MS product? You get back something like $70. A copy of Vista Ultimate OEM from Newegg is $189.99. Of course, I'd imagine that Dell gets an even higher discount.
In fact, Vista Ultimate Retail at Newegg is only $378.99. I'm not sure where the article is getting their other numbers. Even if this is the upgrade price, it's already cheaper than the upgrade price listed in the article.
Regardless, with Vista's new pricing model, the difference in cost is even bigger. The extra cost to buy Ultimate over Home Premium probably isn't justified by the effort required to develop the extra pieces. It's probably pointless to speculate over what each version of the OS "cost" to develop per unit.
How will this affect carbon dating, which relies on a ratio involving the amount of carbon in the atmosphere? Beware that law of unintended consequences.
My experience has really been quite the opposite. When Ubuntu's 7.04 came out, I downloaded it using Bittorrent. My ports were forwarded correctly, yet I couldn't achieve more than about half of my DSL's bandwidth. Thinking that my upstream could be to blame, I tried it at the university. I got the exact same speeds (maybe 75kb/s).
Verizon could be throttling, but I'm well informed as to the network infrastructure at the university, and there were no firewall/throttling problems.
Having an IP is enough to get a search warrant, which is enough to seize the computer in question and examine it. There was a story about this not too long ago on Slashdot, but it was mislabeled.. something about open access points not being enough to prevent a conviction (completely inaccurate, but sensational topic).
That can be addressed, too, by requiring proxies to access actual Internet services. Don't like it? Move off-campus and get your own connection, where we won't have to deal with the complaints.
Since the problem is typically providing the copyrighted material (rather than simply downloading it), this would solve a lot of those problems. People would bitch, and the university would point to the p2p problem and explain that it was their fellow students who caused the lockdown.
The whole thing irritates me, but there doesn't appear to be an end in sight. You're corresponding with someone who deals with this problem daily, and with 8-10 complaints per day, it's a pain in the ass. I'd love a solution that doesn't involve restricting the students, allows us to maintain reasonably long logs, and doesn't cause our bandwidth to spike all day long. Right now, my best idea is to outsource the dorm network. I don't think my bosses will go for that, though.
Exactly. I bet the same people complaining that that "isn't a good reason" would be up-in-arms about a tuitition increase intended to address the issue (by hiring more staff, cleaning up the infrastructure to make dealing with the complaints easier, etc).
The single biggest excuse I hear for not blocking e.g. Bit Torrent, is that "I download Linux ISOs!" Yeah. That's a great battle cry, because it was one of the alleged intended purposes for the protocol, but I doubt there's even a single distro which distributes exclusively over BT, and even if there is, it's got to be one of the tiniest niche distros around. And don't get me started on pure p2p networks, like the old Gnutella, KaZaA, eDonkey.
First of all, accounting is a pain in the ass. Billing is problematic, and tacking it onto the fee statement tends to get irate parents calling once the bill comes due. "What do you mean Johnny racked up $1000 in bandwidth usage this semester?!"
Also, dealing with copyright complaints is time-consuming. The requirements in dealing with these notices include not only determining the name of the user who allegedly infringed, but also removing the infringing content. In the case of a university network, this means contacting the student regarding the incident, and telling them to stop sharing.
Blocking en masse means that word will spread quickly. You'll get a lot of complaints at first, but they'll trickle off once it becomes the norm. I don't necessarily agree with their decision, but it certainly means less work for the university staff in the long run.
Chances are most of us will reach for our favorite Linux distro. I'm gonna be that asshole who says, "Not me!" When I need to deploy a system, I take several things into account: maintainability and security are two of the biggest ones. Performance is also a big concern.
I consider Linux best suited as an alternate desktop machine. I tend to use it solely for this purpose. Most of my servers are FreeBSD, because I like how clean and simple it is, and it tends to perform fairly well. It's also a snap to maintain, though not always as easy as some binary distros of Linux (FreeBSD finally got a good binary security update system, though!) I reject OpenBSD outright because the upgrade path is atrocious, and support is fairly limited for older versions. NetBSD certainly seems fine enough, but I've never seen anything it particularly excels at.
I stay away from Solaris for mostly the same reasons that I stay away from OpenBSD--maintainability. That said, if I was trying to serve filesystems from a general purpose box, you can bet I'd be using Solaris and ZFS (which is available in FreeBSD current, but not as stable as Solaris).
But here's the kicker: it would be pretty nice if all of these systems were good at all of these tasks. That is, it'd be nice if there was an OS called SoLinBSD, which had the maintainability and simplicity of FreeBSD, security of OpenBSD, software binary updates (and desktop capabilities) of Linux, and a good ZFS implementation. It would be nice if we could have a homogeneous OS to ease administration. It'd be nice.
The Windows monoculture does this. As the article points out, it actually does a lot of things pretty well, and the interfaces are pretty well documented. It doesn't do everything I want, which is why I don't use it for these tasks, but some people consider a generic tool which does lots of things in an acceptable way to be superior to lots of separate, exceptional tools. That's why we have general purpose computers instead of standalone word processors, e-mail readers, web portals, and game consoles. It's why firewalls are often run on generic hardware when extremely high bandwidth doesn't require an embedded solution. We have a virtual monoculture of hardware (even Macs have gone to x86) for most purposes.
I didn't realize that anyone was advocating the removal of choice. I thought the point was an explanation of how the monoculture has helped Microsoft to succeed.
In fact, it's impossible to take choice away from FOSS, because you can always fork and start a new project. I suppose some sort of TPM which prevented Linux from booting could be an issue, but that's certainly nothing I'd suggest, either.
As for Google--I'm talking about what it is today. It doesn't have the biggest market share, outside of search. Imagine if there was no Yahoo mail, no hushmail, no other medium-sized mail services. If Google had more of a shot at grabbing those users, they would be able to compete more with MSN/Hotmail/whatever.
The difference is that we can force evolution in computers. If a virus comes along, we can detect it. If a vulnerability is discovered, we can patch it. There is no such thing as a computer virus which is incurable and irreparably destroys any computer it touches.
In fact, while a monoculture means that you're more susceptible to the virus, in the computer world, it means that it's easier to fix. A buffer overrun in Mozilla may have to be addressed in several different places in the code (depending upon whether it was in a portable section of code, or if the same techniques were used in sections compiled specifically for Linux/MS/OS X/BeOS/Windows Moble/whatever). In a monoculture, this is less likely.
Do you know someone who's actually done Linux support to verify these unfounded fears? Yes. I've known plenty of people who supported a number of Linux users in a non-homogeneous environment. Part of the problem was that the people who were familiar with Debian knew apt like the back of their hand, but had problems with RPM. Reverse that for people using a Redhat-based distro. Gentoo baffled many of these people, until someone took the plunge and started messing with it. Now you've got Ubuntu, which prettifies a mostly Debian-back end, but if you go in and change something the old-fashioned way, the GUI stuff sometimes ceases to work.
The one good thing I can say about most pre-Ubuntu Linux users was that they had slightly more knowledge. They knew that they were using an alternate OS, and they typically knew the name of the distro they used, so you didn't waste any time trying to figure out which of the multitude of initialization scripts, package managers, and even (in some cases) default window managers were around. People who switched FROM the norm (both within Linux, in the case of the window manager, and in general, in the case of using Linux at all) tended to know what they were using. That didn't mean that support was easy.
Your buddies complain about 3 versions of Windoze because they don't work together - each has it's own arbitrary and insane limitations of a sort not found in the free software world. Simply untrue. One example is Network-Manager, which can't handle certain wireless NICs because they don't talk to the kernel the using the same interface (and we're not even getting into NDIS stuff, here, which is a completely different issue.) Furthermore:
free software all works together Be careful with this. Linux kernel versions may expose or modify APIs causing breakage. Some open source software targeted at Linux simply does not run on other free and open operating systems, such as FreeBSD or OpenBSD. ABI changes in libraries can cause conflicts--installing both libraries can be difficult (if not impossible), meaning that software which relies on a specific version may be incompatible with software relying on a different version (I've seen software which required me to downgrade my version of libc, for example).
Free does not mean that it will work together. A worst case scenario is that you can make it work by rewriting it or fixing the code, but when we're defining success by comparing the user base, you have to take into account that most people won't be able to manage this sort of thing.
Note that this doesn't make it worse than the Windows issues. The fact is that software changes, and sometimes those changes cause incompatibilities. With the Windows monoculture, however, there's only ever one moving target. With Linux, there can be many.
I've been hearing this FUD about "confusion of choice" for a while. As usual, it's designed for people who've never taken so much as a peak outside their favorite non free OS. Anyone who's run free software for more than six months knows it's nonsense. I've been using Linux as my primary desktop OS for about 5 years now. FreeBSD is my server OS of choice. I haven't had Windows installed on a computer I own in probably 3 years, unless you count through emulation (which I use exclusively for support issues with family and friends). I hold to my opinion that confusion of choice is a legitimate issue. It may not be the fact that there are too many choices so much as the underlying reason that there are so many choices--FOSS developers target multiple platforms, interfaces, kernel/libc versions, etc. Spending the time and effort to actually make one distribution good (i.e. coding with that platform in mind, working on usability issues with that platform, etc) could have a good effect on Linux as a whole. Then again, it's also possible that it would become a "too many cooks" issue.
If there was a community with similar interests but widely varied products (think 10 clones of Google, all with mostly the same services, and each with its own small sub-community) it would likely be harder for all of them to survive against a common competitor than for a single Google with all of the userbase of the sub-commmunities.
We celebrate the diversity of choices available to solve a problem and call it freedom. IT managers and CIOs look at it and call it chaos, confusion and uncertainty. That's a pretty solid difference between the two, and honestly, I think it's probably a good point. Linux desktops compete with other Linux desktops. Gnome competes with KDE. It's still choice, and I still like choice, but fragmentation and a hundred different ways of doing things makes it hard to find the information you're looking for online, makes it hard to support (Helpdesk workers complain about having to support more than 3 versions of Windows!), and makes it hard for the user to choose.
It's said multiple times in each RIAA/MPAA story: civil suits do not have to pass the reasonable doubt test. They only have to pass the preponderance of the evidence test. This is because the reasonable doubt test would rarely even apply in civil cases, and when it DID apply, it would be an almost insurmountable barrier. You'd get nasty people causing damage to other people and getting away with it without compensation to the victim. It's just unfortunate that corporations and individuals are considered to be on the same playing field when it comes to courts.
It's just wrong that corporations should not be able to force artists into contracts which deny them any profits after millions of dollars worth of sales. You want to protect the artists from themselves? They entered into those contracts freely.
You can set up other methods for activating it, though. At least on the Mac. By default, I think, the corners of the screen will activate various functions. On a notebook, this may not be as useful, however with a real mouse, moving into the upper-right corner of the screen is quick and easy enough that it can be faster than ALT+TAB if you have a lot of windows open. Similarly, if you have many types of the same window open (xterms, for me) it's pretty useful because ALT+TAB may not give you immediate information on which term you've selected.
I wouldn't have thought that you'd need to touch a Mac in order to know about it. For awhile, it was one of the newest whiz-bangiest features of OS X. That and Dashboard.
And yeah, I guess that's the point. The really useful features in Beryl are really useful. If you get rid of the eye-candy, it could actually make a person more productive.
I saw someone mention Metisse. I'll have to check that one out, too, but probably not if I'm required to use Mandriva.
You didn't by any chance try using the System > Admin > Networking since then, did you? Doing that will break NM. I couldn't swear that I didn't.
As for the rest.. yeah, the article could have been a bit better. Your blog post is more informative, but this guy just wanted to complain, I think. BTW, in case anyone else tries to read your blog--get rid of the trailing slash in that link:)
Setup was just sudo aptitude install network-manager-gnome and I did comment out the interfaces config file. I'm fairly confident that I messed with NM in late 2006, so that would have been before the 2.6.17-11 update. Perhaps newer versions of Network Manager work better?
Anyway, I'm not really complaining, just noting that it didn't work for me, even though I have a supported card. I manage fine with wpa-supplicant, in general, though when I need to connect to WEP "protected" networks, it fails and I have to go edit my interfaces file to use the built-in setup.
Slashdot Mirror
How are Windows users forced? Is a gun held to their heads? Is it in the contract?
My mother used Windows 98 for years after XP had come out. It worked fine and did what she needed.
If you want to test a new distribution scheme, you can use a lab setup. If you need a lot of nodes, you can use http://www.planet-lab.org/. If 99.9% of the use of the new technology has nothing to do with research, I'm not sure I see the value in letting that use hamper the rest of the research at your institution.
Except that the cost to OEMs is significantly cheaper. Ever get a refund for not using the MS product? You get back something like $70. A copy of Vista Ultimate OEM from Newegg is $189.99. Of course, I'd imagine that Dell gets an even higher discount.
In fact, Vista Ultimate Retail at Newegg is only $378.99. I'm not sure where the article is getting their other numbers. Even if this is the upgrade price, it's already cheaper than the upgrade price listed in the article.
Regardless, with Vista's new pricing model, the difference in cost is even bigger. The extra cost to buy Ultimate over Home Premium probably isn't justified by the effort required to develop the extra pieces. It's probably pointless to speculate over what each version of the OS "cost" to develop per unit.
What was it that made Ultimate worth a damn over Premium, again?
How will this affect carbon dating, which relies on a ratio involving the amount of carbon in the atmosphere? Beware that law of unintended consequences.
My experience has really been quite the opposite. When Ubuntu's 7.04 came out, I downloaded it using Bittorrent. My ports were forwarded correctly, yet I couldn't achieve more than about half of my DSL's bandwidth. Thinking that my upstream could be to blame, I tried it at the university. I got the exact same speeds (maybe 75kb/s).
Verizon could be throttling, but I'm well informed as to the network infrastructure at the university, and there were no firewall/throttling problems.
Having an IP is enough to get a search warrant, which is enough to seize the computer in question and examine it. There was a story about this not too long ago on Slashdot, but it was mislabeled.. something about open access points not being enough to prevent a conviction (completely inaccurate, but sensational topic).
That can be addressed, too, by requiring proxies to access actual Internet services. Don't like it? Move off-campus and get your own connection, where we won't have to deal with the complaints.
Since the problem is typically providing the copyrighted material (rather than simply downloading it), this would solve a lot of those problems. People would bitch, and the university would point to the p2p problem and explain that it was their fellow students who caused the lockdown.
The whole thing irritates me, but there doesn't appear to be an end in sight. You're corresponding with someone who deals with this problem daily, and with 8-10 complaints per day, it's a pain in the ass. I'd love a solution that doesn't involve restricting the students, allows us to maintain reasonably long logs, and doesn't cause our bandwidth to spike all day long. Right now, my best idea is to outsource the dorm network. I don't think my bosses will go for that, though.
Yeah, plus, then when people start doing really bad things like child porn or e-threats, we'll be protecting them, too!
Besides, it won't matter once ISPs get a legal requirement to maintain records. It's coming, don't doubt it.
Exactly. I bet the same people complaining that that "isn't a good reason" would be up-in-arms about a tuitition increase intended to address the issue (by hiring more staff, cleaning up the infrastructure to make dealing with the complaints easier, etc).
You hit the nail in the coffin on the head.
The single biggest excuse I hear for not blocking e.g. Bit Torrent, is that "I download Linux ISOs!" Yeah. That's a great battle cry, because it was one of the alleged intended purposes for the protocol, but I doubt there's even a single distro which distributes exclusively over BT, and even if there is, it's got to be one of the tiniest niche distros around. And don't get me started on pure p2p networks, like the old Gnutella, KaZaA, eDonkey.
First of all, accounting is a pain in the ass. Billing is problematic, and tacking it onto the fee statement tends to get irate parents calling once the bill comes due. "What do you mean Johnny racked up $1000 in bandwidth usage this semester?!"
Also, dealing with copyright complaints is time-consuming. The requirements in dealing with these notices include not only determining the name of the user who allegedly infringed, but also removing the infringing content. In the case of a university network, this means contacting the student regarding the incident, and telling them to stop sharing.
Blocking en masse means that word will spread quickly. You'll get a lot of complaints at first, but they'll trickle off once it becomes the norm. I don't necessarily agree with their decision, but it certainly means less work for the university staff in the long run.
I consider Linux best suited as an alternate desktop machine. I tend to use it solely for this purpose. Most of my servers are FreeBSD, because I like how clean and simple it is, and it tends to perform fairly well. It's also a snap to maintain, though not always as easy as some binary distros of Linux (FreeBSD finally got a good binary security update system, though!) I reject OpenBSD outright because the upgrade path is atrocious, and support is fairly limited for older versions. NetBSD certainly seems fine enough, but I've never seen anything it particularly excels at.
I stay away from Solaris for mostly the same reasons that I stay away from OpenBSD--maintainability. That said, if I was trying to serve filesystems from a general purpose box, you can bet I'd be using Solaris and ZFS (which is available in FreeBSD current, but not as stable as Solaris).
But here's the kicker: it would be pretty nice if all of these systems were good at all of these tasks. That is, it'd be nice if there was an OS called SoLinBSD, which had the maintainability and simplicity of FreeBSD, security of OpenBSD, software binary updates (and desktop capabilities) of Linux, and a good ZFS implementation. It would be nice if we could have a homogeneous OS to ease administration. It'd be nice.
The Windows monoculture does this. As the article points out, it actually does a lot of things pretty well, and the interfaces are pretty well documented. It doesn't do everything I want, which is why I don't use it for these tasks, but some people consider a generic tool which does lots of things in an acceptable way to be superior to lots of separate, exceptional tools. That's why we have general purpose computers instead of standalone word processors, e-mail readers, web portals, and game consoles. It's why firewalls are often run on generic hardware when extremely high bandwidth doesn't require an embedded solution. We have a virtual monoculture of hardware (even Macs have gone to x86) for most purposes.
Ug. Rant off.
I didn't realize that anyone was advocating the removal of choice. I thought the point was an explanation of how the monoculture has helped Microsoft to succeed.
In fact, it's impossible to take choice away from FOSS, because you can always fork and start a new project. I suppose some sort of TPM which prevented Linux from booting could be an issue, but that's certainly nothing I'd suggest, either.
As for Google--I'm talking about what it is today. It doesn't have the biggest market share, outside of search. Imagine if there was no Yahoo mail, no hushmail, no other medium-sized mail services. If Google had more of a shot at grabbing those users, they would be able to compete more with MSN/Hotmail/whatever.
The difference is that we can force evolution in computers. If a virus comes along, we can detect it. If a vulnerability is discovered, we can patch it. There is no such thing as a computer virus which is incurable and irreparably destroys any computer it touches.
In fact, while a monoculture means that you're more susceptible to the virus, in the computer world, it means that it's easier to fix. A buffer overrun in Mozilla may have to be addressed in several different places in the code (depending upon whether it was in a portable section of code, or if the same techniques were used in sections compiled specifically for Linux/MS/OS X/BeOS/Windows Moble/whatever). In a monoculture, this is less likely.
The one good thing I can say about most pre-Ubuntu Linux users was that they had slightly more knowledge. They knew that they were using an alternate OS, and they typically knew the name of the distro they used, so you didn't waste any time trying to figure out which of the multitude of initialization scripts, package managers, and even (in some cases) default window managers were around. People who switched FROM the norm (both within Linux, in the case of the window manager, and in general, in the case of using Linux at all) tended to know what they were using. That didn't mean that support was easy. Your buddies complain about 3 versions of Windoze because they don't work together - each has it's own arbitrary and insane limitations of a sort not found in the free software world. Simply untrue. One example is Network-Manager, which can't handle certain wireless NICs because they don't talk to the kernel the using the same interface (and we're not even getting into NDIS stuff, here, which is a completely different issue.) Furthermore: free software all works together Be careful with this. Linux kernel versions may expose or modify APIs causing breakage. Some open source software targeted at Linux simply does not run on other free and open operating systems, such as FreeBSD or OpenBSD. ABI changes in libraries can cause conflicts--installing both libraries can be difficult (if not impossible), meaning that software which relies on a specific version may be incompatible with software relying on a different version (I've seen software which required me to downgrade my version of libc, for example).
Free does not mean that it will work together. A worst case scenario is that you can make it work by rewriting it or fixing the code, but when we're defining success by comparing the user base, you have to take into account that most people won't be able to manage this sort of thing.
Note that this doesn't make it worse than the Windows issues. The fact is that software changes, and sometimes those changes cause incompatibilities. With the Windows monoculture, however, there's only ever one moving target. With Linux, there can be many. I've been hearing this FUD about "confusion of choice" for a while. As usual, it's designed for people who've never taken so much as a peak outside their favorite non free OS. Anyone who's run free software for more than six months knows it's nonsense. I've been using Linux as my primary desktop OS for about 5 years now. FreeBSD is my server OS of choice. I haven't had Windows installed on a computer I own in probably 3 years, unless you count through emulation (which I use exclusively for support issues with family and friends). I hold to my opinion that confusion of choice is a legitimate issue. It may not be the fact that there are too many choices so much as the underlying reason that there are so many choices--FOSS developers target multiple platforms, interfaces, kernel/libc versions, etc. Spending the time and effort to actually make one distribution good (i.e. coding with that platform in mind, working on usability issues with that platform, etc) could have a good effect on Linux as a whole. Then again, it's also possible that it would become a "too many cooks" issue.
If there was a community with similar interests but widely varied products (think 10 clones of Google, all with mostly the same services, and each with its own small sub-community) it would likely be harder for all of them to survive against a common competitor than for a single Google with all of the userbase of the sub-commmunities.
It's said multiple times in each RIAA/MPAA story: civil suits do not have to pass the reasonable doubt test. They only have to pass the preponderance of the evidence test. This is because the reasonable doubt test would rarely even apply in civil cases, and when it DID apply, it would be an almost insurmountable barrier. You'd get nasty people causing damage to other people and getting away with it without compensation to the victim. It's just unfortunate that corporations and individuals are considered to be on the same playing field when it comes to courts.
You can set up other methods for activating it, though. At least on the Mac. By default, I think, the corners of the screen will activate various functions. On a notebook, this may not be as useful, however with a real mouse, moving into the upper-right corner of the screen is quick and easy enough that it can be faster than ALT+TAB if you have a lot of windows open. Similarly, if you have many types of the same window open (xterms, for me) it's pretty useful because ALT+TAB may not give you immediate information on which term you've selected.
Well, http://en.wikipedia.org/wiki/Expos%C3%A9_(Mac_OS_X ), but I was too lazy to find out how to add the accent.
I wouldn't have thought that you'd need to touch a Mac in order to know about it. For awhile, it was one of the newest whiz-bangiest features of OS X. That and Dashboard.
And yeah, I guess that's the point. The really useful features in Beryl are really useful. If you get rid of the eye-candy, it could actually make a person more productive.
I saw someone mention Metisse. I'll have to check that one out, too, but probably not if I'm required to use Mandriva.
It seems to be blogger.com that didn't like it. Gave me a 404 in Opera and Firefox.
As for the rest.. yeah, the article could have been a bit better. Your blog post is more informative, but this guy just wanted to complain, I think. BTW, in case anyone else tries to read your blog--get rid of the trailing slash in that link
Setup was just sudo aptitude install network-manager-gnome and I did comment out the interfaces config file. I'm fairly confident that I messed with NM in late 2006, so that would have been before the 2.6.17-11 update. Perhaps newer versions of Network Manager work better?
Anyway, I'm not really complaining, just noting that it didn't work for me, even though I have a supported card. I manage fine with wpa-supplicant, in general, though when I need to connect to WEP "protected" networks, it fails and I have to go edit my interfaces file to use the built-in setup.