> And as for the multimedia terminal, I'd personally rather have one machine that does it all
Why? So you have to run RG6 to all the media terminals as well, and have extra hard drives all over the house? I'd much rather have one or two central machines beefed up with all the storage they can take and multiple TV tuners, serving up video to thin client media terminals throughout the house. A $99 MythTV client based on the GC for each TV in the house would be just the ticket. Eventually after they're EOLed you'd hope to pick them up for well under $100.
These are probably the same people that notice that you can turn in a swivel chair by swinging your outstreched arms quickly in one direction and slowly in the other. Sadly not all of them experience the aha-moment of learning about bearing friction.
> What do you do if your robotic arm loses a necessary screw?
And what do you do if your orbiter loses a necessary patch of insulation? Sure, humans in orbit give you added flexibility, at much added cost. Overall the loss of flexibility with purely robotic maintenance could be offset with extra backup mechanisms and some more up-front design thought, plus the cheaper cost of not having humans there.
You first said "very internationaL" effort. Considering all the tech is provided by two German companies, at best it's a two-country effort. How is that "very" international? Or maybe you want to explain what you meant by "very".
What a load of crap. If anything, the Shanghai project has cost German tax payers money, since the German government essentially subsidized it. Der Spiegel has always hated the Transrapid from day one, and they never miss a chance to diss it. Saying that the German economy depends on the technology of two of its companies is a bit rich even for Der Spiegel, though.
It's ironic how much the Greens have hated the Transrapid, for reasons only they know. Probably because it's high-tech, and Greens deep down are but simple luddites. First it was the noise, and when independent data showed that the Transrapid is actually considerably quieter than conventional trains, it because the energy usage. When that was struck down also, the arguments became more and more bizzare. If anything, the Greens should embrace the Transrapid. It is much cleaner at the point of use (no oil dripping along the track like conventional trains), quieter at high speed and practically silent at low speed in urban areas, the track uses MUCH less real estate (it could even be stacked in tight urban areas) and can be integrated into the environment much more benignly with tighter curves and steeper grades--IOW less terraforming would be required.
Or it could be a strategy to squeeze the price even lower. The Chinese already strong-armed Thyssen and Siemens into installing the Shanghai system at zero profit. They agreed because they wanted a show piece for the technology, hoping for future sales. But I doubt they'll go for a mammoth project of over 1000 km without making a dime. They couldn't afford to.
Well, not really. The Transrapid, which wraps around the rail, could be seriously damaged and even derailed by a concrete or steel wedge placed with the thin edge facing the oncoming train. Attempts and actual derailments have already taken place on the German ICE using concrete blocks.
> any large-scale engineering effort of these sorts of things are usually a -very- international effort
Not in this case. Most patents are held by Thyssen and Siemens, both of which are German companies.
> Flippantly assuming that just because the Chinese are the 'Bad Guys' they'll > ignore all business regulation, well... thats just a tad ignorant my friend, > and extremely blissless.
Or hopelessly starry-eyed in your case. The Chinese are already under strong suspicion of having hijacked much of the Transrapid technology to advance their own homegrown maglev efforts. Within a short time of starting the Transrapid contract they announced major "breakthroughs" in their own research. Furthermore the Thyssen engineers supervising in Shanghai reported that they were often denied access to local fabrication plants where Transrapid components were manufactured under license. One of the conditions of the contract was that the Chinese would be allowed to manufacture their own track under license. Considering that with a maglev the track is a very high-tech and important part of the complete system (the other being the feedback levitation system on the train itself), this is a huge concession.
So yes, I definitely wouldn't put very large-scale industrial espionage past the Chinese. Or any other country, for that matter.
when I first read the headline "Wireless Street Lamps for Traffic Monitoring" that the UK were sinking yet more money in that elusive pseudo-tech called "wireless light" that would enable police to see cars even at night. Thank goodness it turned out to be something much more sensible.
In other news, recent DNA tests determined that George Orwell was in fact a filthy Kraut masquerading as an English subject to perpetrate immoral thought crimes onto this noble people.
> two TGV trains crossed each others at a relative speed of 777 km/h
Ouch!!! For the next test, and with a fresh supply of trains and people, they decided to just pass each other instead. Things worked much better that time.
> And if a Transrapid train has to stop on the curve?
It wraps around the track, so it can't tip over even when stationary. Doesn't mean it would be particularly comfortable for the passengers while stationary on such a tilting track, but the train would be fine.
> Nesting ip addresses has the exact same effect as extending the address space we currently have.
No it doesn't. You're talking one global address space versus many private partitioned ones.
> In addition to that, the delivery of your packet is now DEPENDANT on the > external router which forwards the rest of the way internally, right?
Well, it is dependent on it anyway, since it's the only point of connection between the public and private networks, "the internet" won't be able to find an alternate route anyway. But yes, as I said in other posts, this is essentially a hard-coded route.
> More IPs doesn't mean that somehow DNS is going to suffer...
Really? So how do YOU think IP addresses are resolved--the computer just somehow knows, right? Those addresses don't actually need to be stored somewhere, and there's no network traffic involved in getting them to your computer? DNS is already suffering because of poor system and application configurations, it will only get worse once you increase the number of participants by orders of magnitude.
> The problem is exposing standard services from behind the NAT by more than one machine.
Well, that's exactly one of the shortcomings of port forwarding. As I said, ports were originally meant to identify applications on a machine, not machines on a network. If you use them as machine identifiers, you can't also use them as app identifiers on those machines. You just can't have it both ways without some serious hacking (assigning port ranges to various machines etc.)
You're preaching to the choir, but that doesn't help anybody. Instead you have to work with what you have. I'm also not aware of any ISPs that encourage NAT use--quite the opposite actually, I hear of many that are starting to crack down on home NAT use. This is usually in order to sell you multiple IP addresses instead, or--even more ludicrous--multiple cable modems.
> Frankly, I'd never get services with a provider that required me to pay extra for more IP
Must be nice, the world you're living in. Around here (US) broadband providers are a very scarce resource--they own YOU, you don't own them. Where I live I have a choice of cable or DSL, and both providers SELL additional IP addresses.
> it's no better or worse then 1.2/3.4 (just that it's longer).
It's arbitrarily long, since you can always partition off another private network off an existing private network and nest another packet envelope to address it.
> His suggestion still results in the removal of NAT, and replace it with more routers and more address space.
My point is that you don't necessarily WANT all IP-based devices to be public. With IPv6 you have 128 bits of address, so let's assume that for whatever obsure reason I want to create a public LED billboard where each LED has its own IP address. And I sell hundreds of thousands of these all over the world. Now I can send a packet to any one of these LEDs in the world to switch it individually. All these billions of LEDs pollute DNS tables and routers all over the world, not to mention the traffic overhead of sending individual packets to each. The point is that just because you can doesn't mean you should.
> If a private address is reachable, then why is it still private?
[...] make it public and be done with it.
For various reasons. Some people and companies feel that certain things should be scarce and/or expensive, regardless of how plentiful they are. To some a public IP address seems to inherently valuable that it shouldn't just be handed out like that. I don't think we'll see a world where IP addresses can be had for pennies or what have you, even with IPv6. Besides, there IS overhead involved with making an address public, since it must enter the DNS system and will occupy router caches along the way. With 32 bits our current internet routing system works. Make it 128 bits and give every light switch, temperature sensor, door know, LED etc. its own address, and the internet will come to its knees.
Having nested packets allows you to partition a private network into its own internet universe, allowing it to use all the same IP technologies and hardware, yet not pollute the public internet with its existence. Look, the current system of public internet/private intranets all using off-the-shelf TCP/IP works great and is exactly what many people like and want--with the exception of port forwarding hacks and such. Consider a nested packet header to be a "hint" to the relevant router as to which machine on a private intranet to forward the packet to, without encumbering the public internet with the overhead of its existence.
> We already have at least two. > One is IPv6. The other is VPNs
IPv6 doesn't solve the problem of how to reach private addresses, it merely provides tons more public ones to eliminate the need for private ones. Except the lack of public addresses isn't the only reason for the modern use of NAT anymore.
Regarding VPNs, it's an interesting way of bypassing the problem by making you a part of the private network, but you get other problems that way. You obtain an IP address on the destination network, and with home-type setups you're liable to get endless clashes--how many people run their home LAN on the 192.168.0 subnet? When you send a packet to 192.168.0.4, does your local print server get it, or your friend's voice chat program on the remote private network? I haven't looked at this issue with VPNs enough to know if there are easy established solutions.
> Can *you*, as the originator of the packet, specify what machine the packet is > routed to given that there are multiple NATed machines listening in on port 5555
How would "multiple NATed machines" "listen in" on the same port? A router normally forwards packets from ONE port to ONE machine, unless you've hacked yours to broadcast them instead on the LAN.
The reason he doesn't get what I mean is that with current routers and IP you cannot directly send a packet from outside to address 192.168.1.3 on your home LAN without mapping it to a particular port. With a nested IP scheme your router would receive the packet because of the outer-most envelope, and would then examine the inner envelope to find the next destination of 192.168.1.3. Essentially we're talking about explicit routing in a sense.
> I laughed when i read this. What you just described is a router. A plain regular router of which thousands exist.
You don't quite follow me, I'm afraid. In my example, the only public (and publicly routable) IP address is a.b.c.d, which as far as the greater internet is concerned, is the final destination address of this packet, and the payload of the packet is opaque as far as anybody is concerned. In our case it's actually the address of a router. Only this router cares that there is another nested packet inside, with a further destination address, e.f.g.h. This is NOT a public IP address, and is thus not publicly routable. In a home user's case it could be 192.168.0.3, the address of your internet-attached vibrator. Only your router at a.b.c.d knows how to route the packet on to e.f.g.h. Furthermore, let's say your vibrator is in fact a wireless router that services your webcams scattered throughout your "compound". For whatever paranoid reason you don't even want these to be attached to your main intranet directly. So now to address webcam B to see Dee at i.j.k.l, your packet has three nested headers: a.b.c.d/e.f.g.h/i.j.k.l. Only the wireless router at e.f.g.h knows how to route to i.j.k.l, etc. For simplicity and genericity you would make this nesting scheme unlimited, but for most setups it would be a two level system--internet and intranet.
Incidentally, this scheme is hardly innovative. Especially in industrial settings you often find "networks of networks", where gateways connect dissimilar networking technologies that usually don't run IP in a router fashion. The issue is then often how to generically address a device on a sub-sub-network somewhere out there.
It could be useful to extend the IP protocol to allow partitioning off private intranets with machines that don't have to show up in the (ever growing) public DNS tables, yet can still be addressed individually provided you know where you want to go.
> Do you think that there is a coincidence between this, and the canning of > development of a free speech & free beer VOIP product that uses > military-grade encryption?
Actually, the tin-foil-hat part of me did pipe up with this idea while reading the post. But the guy seems to be based in Switzerland, and at least emotionally he would feel less under the auspices of Uncle. Of course, who knows.
Slashdot Mirror
> And as for the multimedia terminal, I'd personally rather have one machine that does it all
Why? So you have to run RG6 to all the media terminals as well, and have extra hard drives all over the house? I'd much rather have one or two central machines beefed up with all the storage they can take and multiple TV tuners, serving up video to thin client media terminals throughout the house. A $99 MythTV client based on the GC for each TV in the house would be just the ticket. Eventually after they're EOLed you'd hope to pick them up for well under $100.
> my gamecube's fan is loud
And gamecube fans are louder still.
These are probably the same people that notice that you can turn in a swivel chair by swinging your outstreched arms quickly in one direction and slowly in the other. Sadly not all of them experience the aha-moment of learning about bearing friction.
> What do you do if your robotic arm loses a necessary screw?
And what do you do if your orbiter loses a necessary patch of insulation? Sure, humans in orbit give you added flexibility, at much added cost. Overall the loss of flexibility with purely robotic maintenance could be offset with extra backup mechanisms and some more up-front design thought, plus the cheaper cost of not having humans there.
You first said "very internationaL" effort. Considering all the tech is provided by two German companies, at best it's a two-country effort. How is that "very" international? Or maybe you want to explain what you meant by "very".
What a load of crap. If anything, the Shanghai project has cost German tax payers money, since the German government essentially subsidized it. Der Spiegel has always hated the Transrapid from day one, and they never miss a chance to diss it. Saying that the German economy depends on the technology of two of its companies is a bit rich even for Der Spiegel, though.
It's ironic how much the Greens have hated the Transrapid, for reasons only they know. Probably because it's high-tech, and Greens deep down are but simple luddites. First it was the noise, and when independent data showed that the Transrapid is actually considerably quieter than conventional trains, it because the energy usage. When that was struck down also, the arguments became more and more bizzare. If anything, the Greens should embrace the Transrapid. It is much cleaner at the point of use (no oil dripping along the track like conventional trains), quieter at high speed and practically silent at low speed in urban areas, the track uses MUCH less real estate (it could even be stacked in tight urban areas) and can be integrated into the environment much more benignly with tighter curves and steeper grades--IOW less terraforming would be required.
Or it could be a strategy to squeeze the price even lower. The Chinese already strong-armed Thyssen and Siemens into installing the Shanghai system at zero profit. They agreed because they wanted a show piece for the technology, hoping for future sales. But I doubt they'll go for a mammoth project of over 1000 km without making a dime. They couldn't afford to.
> 2) It is very difficult to derail
Well, not really. The Transrapid, which wraps around the rail, could be seriously damaged and even derailed by a concrete or steel wedge placed with the thin edge facing the oncoming train. Attempts and actual derailments have already taken place on the German ICE using concrete blocks.
> any large-scale engineering effort of these sorts of things are usually a -very- international effort
... thats just a tad ignorant my friend,
Not in this case. Most patents are held by Thyssen and Siemens, both of which are German companies.
> Flippantly assuming that just because the Chinese are the 'Bad Guys' they'll
> ignore all business regulation, well
> and extremely blissless.
Or hopelessly starry-eyed in your case. The Chinese are already under strong suspicion of having hijacked much of the Transrapid technology to advance their own homegrown maglev efforts. Within a short time of starting the Transrapid contract they announced major "breakthroughs" in their own research. Furthermore the Thyssen engineers supervising in Shanghai reported that they were often denied access to local fabrication plants where Transrapid components were manufactured under license. One of the conditions of the contract was that the Chinese would be allowed to manufacture their own track under license. Considering that with a maglev the track is a very high-tech and important part of the complete system (the other being the feedback levitation system on the train itself), this is a huge concession.
So yes, I definitely wouldn't put very large-scale industrial espionage past the Chinese. Or any other country, for that matter.
when I first read the headline "Wireless Street Lamps for Traffic Monitoring" that the UK were sinking yet more money in that elusive pseudo-tech called "wireless light" that would enable police to see cars even at night. Thank goodness it turned out to be something much more sensible.
> 1984...only 20 years later
In other news, recent DNA tests determined that George Orwell was in fact a filthy Kraut masquerading as an English subject to perpetrate immoral thought crimes onto this noble people.
> Just who do you think gave the world the USA?
Or, more importantly, just who gave the world TO the USA? I'd like to catch that git.
> two TGV trains crossed each others at a relative speed of 777 km/h
Ouch!!! For the next test, and with a fresh supply of trains and people, they decided to just pass each other instead. Things worked much better that time.
> And if a Transrapid train has to stop on the curve?
It wraps around the track, so it can't tip over even when stationary. Doesn't mean it would be particularly comfortable for the passengers while stationary on such a tilting track, but the train would be fine.
> Nesting ip addresses has the exact same effect as extending the address space we currently have.
No it doesn't. You're talking one global address space versus many private partitioned ones.
> In addition to that, the delivery of your packet is now DEPENDANT on the
> external router which forwards the rest of the way internally, right?
Well, it is dependent on it anyway, since it's the only point of connection between the public and private networks, "the internet" won't be able to find an alternate route anyway. But yes, as I said in other posts, this is essentially a hard-coded route.
> More IPs doesn't mean that somehow DNS is going to suffer...
Really? So how do YOU think IP addresses are resolved--the computer just somehow knows, right? Those addresses don't actually need to be stored somewhere, and there's no network traffic involved in getting them to your computer? DNS is already suffering because of poor system and application configurations, it will only get worse once you increase the number of participants by orders of magnitude.
> The problem is exposing standard services from behind the NAT by more than one machine.
Well, that's exactly one of the shortcomings of port forwarding. As I said, ports were originally meant to identify applications on a machine, not machines on a network. If you use them as machine identifiers, you can't also use them as app identifiers on those machines. You just can't have it both ways without some serious hacking (assigning port ranges to various machines etc.)
You're preaching to the choir, but that doesn't help anybody. Instead you have to work with what you have. I'm also not aware of any ISPs that encourage NAT use--quite the opposite actually, I hear of many that are starting to crack down on home NAT use. This is usually in order to sell you multiple IP addresses instead, or--even more ludicrous--multiple cable modems.
> Frankly, I'd never get services with a provider that required me to pay extra for more IP
Must be nice, the world you're living in. Around here (US) broadband providers are a very scarce resource--they own YOU, you don't own them. Where I live I have a choice of cable or DSL, and both providers SELL additional IP addresses.
> it's no better or worse then 1.2/3.4 (just that it's longer).
It's arbitrarily long, since you can always partition off another private network off an existing private network and nest another packet envelope to address it.
> His suggestion still results in the removal of NAT, and replace it with more routers and more address space.
My point is that you don't necessarily WANT all IP-based devices to be public. With IPv6 you have 128 bits of address, so let's assume that for whatever obsure reason I want to create a public LED billboard where each LED has its own IP address. And I sell hundreds of thousands of these all over the world. Now I can send a packet to any one of these LEDs in the world to switch it individually. All these billions of LEDs pollute DNS tables and routers all over the world, not to mention the traffic overhead of sending individual packets to each. The point is that just because you can doesn't mean you should.
> If a private address is reachable, then why is it still private?
[...] make it public and be done with it.
For various reasons. Some people and companies feel that certain things should be scarce and/or expensive, regardless of how plentiful they are. To some a public IP address seems to inherently valuable that it shouldn't just be handed out like that. I don't think we'll see a world where IP addresses can be had for pennies or what have you, even with IPv6. Besides, there IS overhead involved with making an address public, since it must enter the DNS system and will occupy router caches along the way. With 32 bits our current internet routing system works. Make it 128 bits and give every light switch, temperature sensor, door know, LED etc. its own address, and the internet will come to its knees.
Having nested packets allows you to partition a private network into its own internet universe, allowing it to use all the same IP technologies and hardware, yet not pollute the public internet with its existence. Look, the current system of public internet/private intranets all using off-the-shelf TCP/IP works great and is exactly what many people like and want--with the exception of port forwarding hacks and such. Consider a nested packet header to be a "hint" to the relevant router as to which machine on a private intranet to forward the packet to, without encumbering the public internet with the overhead of its existence.
> Nihilism though? I think that's reading into it a bit too much.
Hey, that was conjecture on MY part <g>.
> We already have at least two.
> One is IPv6. The other is VPNs
IPv6 doesn't solve the problem of how to reach private addresses, it merely provides tons more public ones to eliminate the need for private ones. Except the lack of public addresses isn't the only reason for the modern use of NAT anymore.
Regarding VPNs, it's an interesting way of bypassing the problem by making you a part of the private network, but you get other problems that way. You obtain an IP address on the destination network, and with home-type setups you're liable to get endless clashes--how many people run their home LAN on the 192.168.0 subnet? When you send a packet to 192.168.0.4, does your local print server get it, or your friend's voice chat program on the remote private network? I haven't looked at this issue with VPNs enough to know if there are easy established solutions.
> Can *you*, as the originator of the packet, specify what machine the packet is
> routed to given that there are multiple NATed machines listening in on port 5555
How would "multiple NATed machines" "listen in" on the same port? A router normally forwards packets from ONE port to ONE machine, unless you've hacked yours to broadcast them instead on the LAN.
The reason he doesn't get what I mean is that with current routers and IP you cannot directly send a packet from outside to address 192.168.1.3 on your home LAN without mapping it to a particular port. With a nested IP scheme your router would receive the packet because of the outer-most envelope, and would then examine the inner envelope to find the next destination of 192.168.1.3. Essentially we're talking about explicit routing in a sense.
> I laughed when i read this. What you just described is a router. A plain regular router of which thousands exist.
You don't quite follow me, I'm afraid. In my example, the only public (and publicly routable) IP address is a.b.c.d, which as far as the greater internet is concerned, is the final destination address of this packet, and the payload of the packet is opaque as far as anybody is concerned. In our case it's actually the address of a router. Only this router cares that there is another nested packet inside, with a further destination address, e.f.g.h. This is NOT a public IP address, and is thus not publicly routable. In a home user's case it could be 192.168.0.3, the address of your internet-attached vibrator. Only your router at a.b.c.d knows how to route the packet on to e.f.g.h. Furthermore, let's say your vibrator is in fact a wireless router that services your webcams scattered throughout your "compound". For whatever paranoid reason you don't even want these to be attached to your main intranet directly. So now to address webcam B to see Dee at i.j.k.l, your packet has three nested headers: a.b.c.d/e.f.g.h/i.j.k.l. Only the wireless router at e.f.g.h knows how to route to i.j.k.l, etc. For simplicity and genericity you would make this nesting scheme unlimited, but for most setups it would be a two level system--internet and intranet.
Incidentally, this scheme is hardly innovative. Especially in industrial settings you often find "networks of networks", where gateways connect dissimilar networking technologies that usually don't run IP in a router fashion. The issue is then often how to generically address a device on a sub-sub-network somewhere out there.
It could be useful to extend the IP protocol to allow partitioning off private intranets with machines that don't have to show up in the (ever growing) public DNS tables, yet can still be addressed individually provided you know where you want to go.
> Furthermore I propose we call this a 'port'.
Cute, but I already mentioned its existence in my post, and the need for something else.
> Do you think that there is a coincidence between this, and the canning of
> development of a free speech & free beer VOIP product that uses
> military-grade encryption?
Actually, the tin-foil-hat part of me did pipe up with this idea while reading the post. But the guy seems to be based in Switzerland, and at least emotionally he would feel less under the auspices of Uncle. Of course, who knows.