Many people believe that the salient problem with AFS is that it violates unix
semantics. AFS has a program called "fs" that facilitates dealing with
metadata like permissions, etc. For example, chmod doesn't do anything in the
AFS environment; you need "fs sa" instead.
On the other hand, if GFS doesn't do something intelligent about security, then
we're left with the same fundamental problem that NFS has. Namely, we need to
presume that it operates within a local environment in which all users on the
inside are trusted. (Insert end-to-end argument here.)
Obviously the idea of "secure network" is a myth, and when I first glanced at
the headline on Slashdot, I was hoping that GFS would be a step in the right
direction toward a secure filesystem that actually stands a chance of
being implemented in servers like the ones produced by NetApp. I guess I am
disappointed.
Consider that some Internet services allow arbitrary posting of content. This includes websites that allow content uploads, collaborative weblogs, filesharing networks, anonymous FTP services, Wikis, -- heck, even Slashdot. Could I post some child pornography in order to justify blocking access to such services? You bet.
This argument could be used to justify generally outlawing services that allow anonymous upload of information. And I don't think that this is where we want to go.
Let's look at the fundamental motivation for the WTO decision here: the WTO consists of parties interested in extracting capital from the US. Gambling is one of the purest ways of accomplishing this: regardless of whether or not markets are efficient, regardless of whether CAPM is the right economic model or not, casino gamblers are pretty much 100% profit.
Now, let's imagine turning on the flow of funds from wealthy but fiscally naive Americans who gamble to wealthy but fiscally shrewd corporations overseas who run casino outfits. Suddenly parties involved in the WTO decision have a nice way of taking money from the Americans and bringing it to the economies of other nations.
This sounds harsh and perhaps cynical, but I appreciate the presence of beggars on the streets of my hometown, since I know that they are bringing in money to our community from tourists. I don't see how this is any different, except that in this case, we're not just concerned with dimes and nickels. There are people in the US with serious gambling addictions, and the people of the US effectively finance their habits in the form of taxes and insurance: the attitude in the US is that someone has to bail out the gamblers in the event that they go broke. And since this is after all a "civilized" nation, I'm inclined to agree. But the point is that the funds are not coming just from a few rich players with money to burn, but from all US citizens -- and sending it to other nations doesn't help resolve the trade deficit or secure America's economic future.
So I worry about this WTO decision. It's not about the US selectively ignoring the interests of other nations; it's about the US protecting itself from those who will take advantage.
With the US experiencing an ever-widening trade deficit, I cannot help but think that American lawmakers have reason to believe that a well-defined market for intellectual property is the only way to secure America's economic future, and that Microsoft and other closed-source software companies are providing the most promising means of sustaining a flow of money into the US.
Essentially, US exports of tangible goods are in decline, and it seems ludicrous to think that providing a management layer for organizations that actually exist overseas can last forever. Do we actually believe that workers in India will not someday discover that if they managed their own companies, then they would not need the US at all? Perhaps this is a bit of an oversimplification, but I think that the point resonates in the hearts and minds of people concerned about economic sustainability for the US. So we're going to support Microsoft, dammit, because OSS is the way to seal America's fate as the country that contributes only intellectual property to the world and gets essentially no compensation in return.
As a supporter of OSS, this notion frightens me. But I have yet to hear an argument that this is not as serious as I fear.
before we leap into IPv6...
on
The State of IPv6
·
· Score: 2, Interesting
Sure, we all love the idea that IPv6 will empower nations that have not managed to accumulate so much address space, and we all love the idea that we may be able to provide corporations with a reasonable excuse not to deploy NAT boxes.
But, before we rush headlong into support of radical IPv6 transformation, we must consider some of the disadvantages. First, there are the costs of migration. Interoperability with IPv4 is an absolute must, lest we make the same mistake that ISO did when it proposed CLNP/CONP in the same breath. Fortunately for us, hardware developers have already seized the opportunity to build IPv6 into routers, and software developers have already integrated IPv6 into the core of popular operating systems such as Linux, Windows, *BSD, etc. But aren't there are some applications that will break if we migrate right away?
Anyway, perhaps that's not a big deal. I'd say the more serious issue is that fast route lookup is made considerably more difficult with the longer prefixes of IPv6. It is fundamentally harder to build switching technology into routers that can handle the longer prefixes and still preserve existing performance guarantees. So unless we don't mind slowing down the internet a bit, we may want to hang on to IPv4 a little longer. Perhaps there is something that ISPs can do such that they can switch IPv6 on shorter prefixes, but I have not yet seen any proposals...
Perhaps more interestingly, integration with Bluetooth or other wireless communication technologies will open the door for person-to-person communication using this system.
What sorts of things would people transfer to each other? Thoughts better left unsaid? A new form of pirated music?
Well, that's not necessarily the case... people who design switching fabrics can tell you that it is much easier to switch quickly on a 32-bit address than it is to switch quickly on a 128-bit address.
One might be inclined to suggest that we would not necessarily have more work to do since there are not 128 bits of entropy in the decisions necessary to route packets among ISPs. However, the special features of IPv6 that use certain bitranges within the address, combined with the way that addresses have been allocated so far, may in fact guarantee that routers will need a deeper switching fabric to handle the different prefix lengths. This deeper switching fabric most certainly will be slower than it could have been with shorter prefixes, and there is no denying that we will take a bit of a hit from it. Not that I don't believe that the benefits outweigh these costs, however...
Is this, then, the end of an era? By the time that third-generation (anonymizing) P2P filesharing networks emerge on the scene, will people have been so whipped that they will no longer be interested in sharing files at all?
OK, so I re-read the parent and discovered that the author had said "last four options" not "only four options". So it is possible that there are technical options as well as political ones (though presently the web site does not load for me, so I cannot verify this). At any rate, I urge people to be careful not to be forced into saying things that they would not ordinarily say.
I for one do use commercial and proprietary software from time to time. My primary reason for using Linux and not Windows, however, is technical: open or not, Linux suits my needs in a way that Windows does not.
Based upon these options, there is a reason not to take the survey seriously: there is no option to claim that you use Linux simply because it is better software. I can see Microsoft's spin already: "100% of respondants say that they use Linux for political reasons rather than technical ones".
It seems that Microsoft has composed its survey choices very carefully: none of the choices in the multiple-choice questions allow participants to express their true feelings about Linux. Most likely, these survey results will be used to convince investors and IT departments that the primary reasons for choosing Linux are political, and thus there is no reason for them to convince their management that it is worth the effort to migrate.
I'm not so sure about this. There are some cases in which children should be able to access internet resources without their parents' knowledge. Take, for example, cases in which children want to find out more about important topics such as contraception or homosexuality, and their parents are unwilling to allow them access to the information.
Public libraries serve this purpose quite well, but in my state (Massachusetts), there is a bill that seeks to disclose children's library activity. Frankly, I find this bill repugnant.
Even if we accept that large-scale P2P systems emerged primarily as a clever way to evade copyright enforcers, we must accept that the ongoing battles between the masses and the enforcers are not limited to copyright-related matters. Throughout human history, oppressors have sustained their power by maintaining a condition in which the masses could not organize. P2P provides a means by which "leaf nodes" can self-organize; this capability has utility not merely to media lovers interested in sharing songs and movies, but to political dissidents interested in arranging meetings or sharing documents.
The possibilities are endless, really -- but if we accept that it will always be the case that some people are wielding power by preventing the masses from organizing, then there will always be a use for P2P.
I don't really care about security either; I don't really care if somone intercepts this
Fair enough for the interception case, but what about the login case? If anyone can claim to be you, then what value is your AOL screenname as an indication of your identity?
The problem with protocols like this is that sooner or later, people will use them to send important or confidential information. Have you ever had a chat with a close friend or SO about something that you would rather not share with the world? Or, perhaps more relevantly, have you ever had a chat with a close friend or SO about something for which it is really important to you that you know with whom you're talking?
But what makes it the best? IM systems consist of three basic components: the protocols, the service, and the participants. If we look at the protocols, AIM is fundamentally not awesome. I'll use some IM protocols from the 1980s -- years before AIM -- as a basis for comparison. IRC offers much greater flexibility, especially for multi-user chat. Zephyr offers hooks to an out-of-band authentication system (Kerberos) which allows for the possibility that a user can authenticate to the server in a way that is actually somewhat secure. (To my knowledge, AIM does not do this.)
Other more recent protocols like Jabber do, however, and how many people are using those?
Next, the service. AOL servers that speak the AIM protocol have this nasty habit of randomly kicking users from time to time. Maybe this is supposed to be "normal", but the two 1980s protocols I mentioned above don't seem to have this problem.
Finally, we have the participants. This is where AOL wins, hands down. Everyone and her mother and her dog use AIM, and thus, if I must choose only one IM protocol, then this must be the one. Many people are effectively faced with this choice... not everyone uses Gaim or Trillian or whatever IM multiplexers there are out there. It seems that if AOL is better, this is the reason, and, I might add, it's not really a good thing.
It seems to me that part of the concern here is that offering these services will encourage further network effects that lock users into a particular IM system, which includes not only the service but the protocol as well. If competition really is important, then why wouldn't the FCC say something like this to AOL? "If you want to provide this kind of service, then you must use a standard format for delivery so that other service providers will be able to compete." It seems to me that the proprietary and constantly changing nature of AOL's AIM protocol and its clients is a large part of what is restricting choice (and, ultimately, innovation) here.
To be honest, I think universities and similar institutions should keep these records, if only because of the number of times I get portscanned or a flood of Code Red/Nimda scans from University IPs...
Most of the time, portscans from university machines are not initiated by the legitimate users of those machines; more often a portscan is a concomitant effect of the legitimate user's machine being compromised! In these cases, revealing the name of the student would almost certainly be a breach of privacy (unless of course we think it's OK to report a student's poor system administration skills).
That process would probably be sufficient to demonstrate identity in court (especially in a civil case, where the standard's not "beyond reasonable doubt" but "preponderance of probabilities").
While this is almost certainly true, it makes me wonder whether people who want to hurt their enemies will just steal the relevant IP addresses and start sharing copyrighted material. Based upon my exposure to university residential and laboratory networks, in most cases such a coup is entirely possible and also rather difficult to prove afterwards.
It's not complete bullshit, actually. The Massachusetts Society for the Prevention of Cruelty to Animals actually has its own police:
"In the years to come, the MSPCA would establish a full-time Law Enforcement Department consisting of highly trained and professional police officers. These officers now attend a 20-24 week state sponsored police training program, and are commissioned as Special State Police Officers to investigate and enforce Massachusetts animal cruelty laws."
As an MIT alumnus who does not use Kazaa, I can honestly say that I see MIT's reluctance to submit to the RIAA as a point in its favor, and my future donations will be more substantial if MIT continues to fight for the civil liberties of its students.
What is to stop the RIAA from using machines in public libraries, or worse, machines connected to large ISPs with pools of DHCP addresses that could correspond to home users? I'm not convinced that every machine they could possibly use for their "nefarious" purposes must come from a particular set of ranges, which seems to be one of the underlying assumptions inherent to this method.
Encryption doesn't entirely solve the problem, since the whole point of these
systems is, after all, the index. And at some level you can't encrypt that, or
else people won't be able to find the files that they want.
Also, ISPs could potentially use ingress filters to block spoofed addresses.
It seems to me that there are two main requirements that an "anonymous" P2P
filesharing system should meet:
1. Data should pass through the overlay network. If I can see addresses other
than the addresses of my neighbors in the overlay (who are presumably my
friends or at least people who trust me in real life), then I can
systematically gather information about who is sharing what, which is
unreasonable. It seems that Freenet is suited to this.
2. Users should be able to publish their own data on their own machines. Most
users aren't so interested in publishing files that they care enough to push it
to someone else's machine, and furthermore, if someone abuses the network it is
a hassle to figure out who to cut off for overpublishing.
3. Performance should be reasonable. Perhaps we can use an approach like
bittorrent or WASTE to exploit parallel downloads, which will probably be
important if we are passing the data through the overlay. We may be able to
use a path-vector algorithm (similar to BGP) to advertise a randomly-generated
node ID through the network, and later on publish index files that associate
parts of a file with the node ID. Not sure if this sort of approach is most
efficient, but it's hard to think of what else we can do.
The bottom line is that we have to avoid connections from nonneighbors, as it
is this that compromises anonymity. We could look to projects like
mixmaster/mixminion or onion routing for anonymity, but I don't think that they
will provide us with the performance we need to keep users happy. Probably the
best thing to do is just route data through your neighbors, but only reveal
your identity to your neighbors -- and no one else.
Slashdot Mirror
On the other hand, if GFS doesn't do something intelligent about security, then we're left with the same fundamental problem that NFS has. Namely, we need to presume that it operates within a local environment in which all users on the inside are trusted. (Insert end-to-end argument here.)
Obviously the idea of "secure network" is a myth, and when I first glanced at the headline on Slashdot, I was hoping that GFS would be a step in the right direction toward a secure filesystem that actually stands a chance of being implemented in servers like the ones produced by NetApp. I guess I am disappointed.
Consider that some Internet services allow arbitrary posting of content. This includes websites that allow content uploads, collaborative weblogs, filesharing networks, anonymous FTP services, Wikis, -- heck, even Slashdot. Could I post some child pornography in order to justify blocking access to such services? You bet. This argument could be used to justify generally outlawing services that allow anonymous upload of information. And I don't think that this is where we want to go.
Now, let's imagine turning on the flow of funds from wealthy but fiscally naive Americans who gamble to wealthy but fiscally shrewd corporations overseas who run casino outfits. Suddenly parties involved in the WTO decision have a nice way of taking money from the Americans and bringing it to the economies of other nations.
This sounds harsh and perhaps cynical, but I appreciate the presence of beggars on the streets of my hometown, since I know that they are bringing in money to our community from tourists. I don't see how this is any different, except that in this case, we're not just concerned with dimes and nickels. There are people in the US with serious gambling addictions, and the people of the US effectively finance their habits in the form of taxes and insurance: the attitude in the US is that someone has to bail out the gamblers in the event that they go broke. And since this is after all a "civilized" nation, I'm inclined to agree. But the point is that the funds are not coming just from a few rich players with money to burn, but from all US citizens -- and sending it to other nations doesn't help resolve the trade deficit or secure America's economic future.
So I worry about this WTO decision. It's not about the US selectively ignoring the interests of other nations; it's about the US protecting itself from those who will take advantage.
Essentially, US exports of tangible goods are in decline, and it seems ludicrous to think that providing a management layer for organizations that actually exist overseas can last forever. Do we actually believe that workers in India will not someday discover that if they managed their own companies, then they would not need the US at all? Perhaps this is a bit of an oversimplification, but I think that the point resonates in the hearts and minds of people concerned about economic sustainability for the US. So we're going to support Microsoft, dammit, because OSS is the way to seal America's fate as the country that contributes only intellectual property to the world and gets essentially no compensation in return.
As a supporter of OSS, this notion frightens me. But I have yet to hear an argument that this is not as serious as I fear.
But, before we rush headlong into support of radical IPv6 transformation, we must consider some of the disadvantages. First, there are the costs of migration. Interoperability with IPv4 is an absolute must, lest we make the same mistake that ISO did when it proposed CLNP/CONP in the same breath. Fortunately for us, hardware developers have already seized the opportunity to build IPv6 into routers, and software developers have already integrated IPv6 into the core of popular operating systems such as Linux, Windows, *BSD, etc. But aren't there are some applications that will break if we migrate right away?
Anyway, perhaps that's not a big deal. I'd say the more serious issue is that fast route lookup is made considerably more difficult with the longer prefixes of IPv6. It is fundamentally harder to build switching technology into routers that can handle the longer prefixes and still preserve existing performance guarantees. So unless we don't mind slowing down the internet a bit, we may want to hang on to IPv4 a little longer. Perhaps there is something that ISPs can do such that they can switch IPv6 on shorter prefixes, but I have not yet seen any proposals...
What sorts of things would people transfer to each other? Thoughts better left unsaid? A new form of pirated music?
One might be inclined to suggest that we would not necessarily have more work to do since there are not 128 bits of entropy in the decisions necessary to route packets among ISPs. However, the special features of IPv6 that use certain bitranges within the address, combined with the way that addresses have been allocated so far, may in fact guarantee that routers will need a deeper switching fabric to handle the different prefix lengths. This deeper switching fabric most certainly will be slower than it could have been with shorter prefixes, and there is no denying that we will take a bit of a hit from it. Not that I don't believe that the benefits outweigh these costs, however...
Is this, then, the end of an era? By the time that third-generation (anonymizing) P2P filesharing networks emerge on the scene, will people have been so whipped that they will no longer be interested in sharing files at all?
I for one do use commercial and proprietary software from time to time. My primary reason for using Linux and not Windows, however, is technical: open or not, Linux suits my needs in a way that Windows does not.
It seems that Microsoft has composed its survey choices very carefully: none of the choices in the multiple-choice questions allow participants to express their true feelings about Linux. Most likely, these survey results will be used to convince investors and IT departments that the primary reasons for choosing Linux are political, and thus there is no reason for them to convince their management that it is worth the effort to migrate.
Isn't Microsoft going to run into trouble for neglecting to enforce its patent early on?
Public libraries serve this purpose quite well, but in my state (Massachusetts), there is a bill that seeks to disclose children's library activity. Frankly, I find this bill repugnant.
The possibilities are endless, really -- but if we accept that it will always be the case that some people are wielding power by preventing the masses from organizing, then there will always be a use for P2P.
Fair enough for the interception case, but what about the login case? If anyone can claim to be you, then what value is your AOL screenname as an indication of your identity?
The problem with protocols like this is that sooner or later, people will use them to send important or confidential information. Have you ever had a chat with a close friend or SO about something that you would rather not share with the world? Or, perhaps more relevantly, have you ever had a chat with a close friend or SO about something for which it is really important to you that you know with whom you're talking?
Next, the service. AOL servers that speak the AIM protocol have this nasty habit of randomly kicking users from time to time. Maybe this is supposed to be "normal", but the two 1980s protocols I mentioned above don't seem to have this problem.
Finally, we have the participants. This is where AOL wins, hands down. Everyone and her mother and her dog use AIM, and thus, if I must choose only one IM protocol, then this must be the one. Many people are effectively faced with this choice... not everyone uses Gaim or Trillian or whatever IM multiplexers there are out there. It seems that if AOL is better, this is the reason, and, I might add, it's not really a good thing.
It seems to me that part of the concern here is that offering these services will encourage further network effects that lock users into a particular IM system, which includes not only the service but the protocol as well. If competition really is important, then why wouldn't the FCC say something like this to AOL? "If you want to provide this kind of service, then you must use a standard format for delivery so that other service providers will be able to compete." It seems to me that the proprietary and constantly changing nature of AOL's AIM protocol and its clients is a large part of what is restricting choice (and, ultimately, innovation) here.
Most of the time, portscans from university machines are not initiated by the legitimate users of those machines; more often a portscan is a concomitant effect of the legitimate user's machine being compromised! In these cases, revealing the name of the student would almost certainly be a breach of privacy (unless of course we think it's OK to report a student's poor system administration skills).
That process would probably be sufficient to demonstrate identity in court (especially in a civil case, where the standard's not "beyond reasonable doubt" but "preponderance of probabilities"). While this is almost certainly true, it makes me wonder whether people who want to hurt their enemies will just steal the relevant IP addresses and start sharing copyrighted material. Based upon my exposure to university residential and laboratory networks, in most cases such a coup is entirely possible and also rather difficult to prove afterwards.
"In the years to come, the MSPCA would establish a full-time Law Enforcement Department consisting of highly trained and professional police officers. These officers now attend a 20-24 week state sponsored police training program, and are commissioned as Special State Police Officers to investigate and enforce Massachusetts animal cruelty laws."
Source: http://www.mspca.org/
As an MIT alumnus who does not use Kazaa, I can honestly say that I see MIT's reluctance to submit to the RIAA as a point in its favor, and my future donations will be more substantial if MIT continues to fight for the civil liberties of its students.
What is to stop the RIAA from using machines in public libraries, or worse, machines connected to large ISPs with pools of DHCP addresses that could correspond to home users? I'm not convinced that every machine they could possibly use for their "nefarious" purposes must come from a particular set of ranges, which seems to be one of the underlying assumptions inherent to this method.
Encryption doesn't entirely solve the problem, since the whole point of these systems is, after all, the index. And at some level you can't encrypt that, or else people won't be able to find the files that they want.
Also, ISPs could potentially use ingress filters to block spoofed addresses. It seems to me that there are two main requirements that an "anonymous" P2P filesharing system should meet: 1. Data should pass through the overlay network. If I can see addresses other than the addresses of my neighbors in the overlay (who are presumably my friends or at least people who trust me in real life), then I can systematically gather information about who is sharing what, which is unreasonable. It seems that Freenet is suited to this. 2. Users should be able to publish their own data on their own machines. Most users aren't so interested in publishing files that they care enough to push it to someone else's machine, and furthermore, if someone abuses the network it is a hassle to figure out who to cut off for overpublishing. 3. Performance should be reasonable. Perhaps we can use an approach like bittorrent or WASTE to exploit parallel downloads, which will probably be important if we are passing the data through the overlay. We may be able to use a path-vector algorithm (similar to BGP) to advertise a randomly-generated node ID through the network, and later on publish index files that associate parts of a file with the node ID. Not sure if this sort of approach is most efficient, but it's hard to think of what else we can do. The bottom line is that we have to avoid connections from nonneighbors, as it is this that compromises anonymity. We could look to projects like mixmaster/mixminion or onion routing for anonymity, but I don't think that they will provide us with the performance we need to keep users happy. Probably the best thing to do is just route data through your neighbors, but only reveal your identity to your neighbors -- and no one else.