Sunday, February 20, 2011

Cerf: Future of Internet doesn't include an IPv7


Vint Cerf takes his title of Chief Internet Evangelist for Google seriously, and is knee-deep in several projects to bring the next versions of the Internet into being. These projects include pushing for worldwide IPv6 adoption, but they don't include plans for an IPv7.
Cerf sat down with Network World's Cisco Subnet editor, Julie Bort, at the annual Digital Broadband Migration conference in Boulder, Colo., to discuss the future of IP, home networking, the Internet of Things, preventing the so-called Internet "kill switch," and other topics. Here is part one of the edited interview.

FULL INTERVIEW: Podcast and transcript
Is there, or will there be, an IPv7, and what new problems will it solve?
During the period of time when we were trying to figure out what to do about expanding in the address space, there were actually four proposals made. Those were eventually narrowed down to one. So actually, 7, 8, 9, don't exist. At the moment there doesn't seem to be any incentive for inventing yet another one. Now the whole problem is to get IPv6 distributed and widely implemented before we literally run out of IPv4 address space
But that shouldn't stop anyone from thinking about new ways of redesigning the Internet. In fact there's a "clean sheet" effort taking place at Stanford University, supported by the National Science Foundation, to look at how would we design the Internet today knowing all the things that we already know about it. At least one concrete outcome has been an "openflow router" which handles packets in a different way than traditional routers. But as far as I know there aren't any new plans for IP-level protocol changes. Someday there will be, I'm sure. (See story: "Showtime for OpenFlow at GENI advanced networking project event")
Will the public/average consumer ever need to make sure all of their home devices (routers, etc.) speak IPv6 ... or will NAT and carriers' efforts to bridge IPv4 to IPv6 be enough for the coming decade?
No, it won't be enough, and I am not a big fan of carrier-grade network address translation. Part of the reason is the whole notion of network address translation is brittle, and it doesn't permit servers to be available on the consumer premises.
In the days when the Internet is highly asymmetric, where you can download traffic faster than you can upload it, putting a server at home is something the broadband providers don't like very much because it consumes a lot of uplink broadband capacity. But with the passage of time, I believe it will be not only desirable, but quite natural, to have servers at home in addition to having use of the cloud. So symmetric capacity and IPv6 are a very attractive outcome and I'm not a big fan of carrier-grade NATs -- but it may turn out that NATs are needed in order to facilitate the transition during this period when we have to run both protocols. (See story: "Can Large Scale NAT Save IPv4?")
Do you think IPv6 in the home is as urgent as it is in the business network?
I think it's urgent in that if we don't get both protocols running at the same time, the day may come when there are servers that can only run IPv6, or there may be users who can only run IPv6 and couldn't get anything else, because the NAT boxes ran out.
I think it's important to get both protocols running smoothly at home. Already laptops and desktops have the capability. It's usually the firewall, the NAT box and maybe the broadband modem that you have at home that haven't been configured for IPv6. (See story: "Cisco Linksys routers still don't support IPv6")
So when we turn on IPv6 on a worldwide basis on June 8 as a 24-hour test (World IPv6 Day), I'm sure there will be things that don't work and those need to be addressed (no pun intended).
I would much rather see a concerted effort to get everybody up and running on IPv6. Then the transition is smooth because it doesn't matter if the destination is running IPv4 or 6, everyone can talk to everyone.
What do you see as the next major consumer use for the Internet?
If I knew that, I would be investing in it. But I think it's pretty clear that social networking is a big part of that. The platforms that are available that allow people to create their own content and share it is probably the most important avenue. The next one will be this Internet of Thingswhere we start managing collections of devices for our benefit. At some point third parties will figure out that they can come in and manage stuff for us. Our entertainment systems might be managed by a third party who doesn't have anything to do with the content, and simply provides organizing capacity, so we can make sure all of our entertainment material shows up on all the platforms where we want it.
You've publicly discussed your own home sensor network that alerts you when your wine cellar gets too hot or dry. You said it was built using technology from Arch Rock, acquired by Cisco in September. Is it Zigbee or Z-wave or something else?
It's actually 802.15.4, 6LoWPAN. It's a 6LoWPAN network running on IPv6.
The recent events in Egypt and the proposed U.S. government "Internet kill switch"... I get the sense that giving government the ability to turn the Internet off this is not something you would approve of. Is there a technological solution to prevent this?
Remember, the Internet runs on a substrate, and if the substrate is controllable by the government, it's possible to turn off the Internet. If we ever get to the point where mesh networking and let's say, peer-wise, point-to-point interactions can be done without benefit of things like routers provided by Internet service providers ... if you can build pieces of the Internet that sort of self-assemble [then the Internet could not be turned off].
Governments will control what they can control if they want to, and so the only solution around that is to have a network that is self-organizing. And technology is available to do that.
Julie Bort is the decided earthbound community editor for Network World. She pens the Odds and Ends blog for Cisco Subnet, the Microsoft Update blog for Microsoft Subnet and the Source Seeker blog for Open Source Subnet. Follow Julie on Twitter @Julie188
Read more about lans & wans in Network World's LANs & WANs section.Is there, or will there be, an IPv7, and what new problems will it solve?
During the period of time when we were trying to figure out what to do about expanding in the address space, there were actually four proposals made. Those were eventually narrowed down to one. So actually, 7, 8, 9, don't exist. At the moment there doesn't seem to be any incentive for inventing yet another one. Now the whole problem is to get IPv6 distributed and widely implemented before we literally run out of IPv4 address space
But that shouldn't stop anyone from thinking about new ways of redesigning the Internet. In fact there's a "clean sheet" effort taking place at Stanford University, supported by the National Science Foundation, to look at how would we design the Internet today knowing all the things that we already know about it. At least one concrete outcome has been an "openflow router" which handles packets in a different way than traditional routers. But as far as I know there aren't any new plans for IP-level protocol changes. Someday there will be, I'm sure. (See story: "Showtime for OpenFlow at GENI advanced networking project event")
Will the public/average consumer ever need to make sure all of their home devices (routers, etc.) speak IPv6 ... or will NAT and carriers' efforts to bridge IPv4 to IPv6 be enough for the coming decade?
No, it won't be enough, and I am not a big fan of carrier-grade network address translation. Part of the reason is the whole notion of network address translation is brittle, and it doesn't permit servers to be available on the consumer premises.
In the days when the Internet is highly asymmetric, where you can download traffic faster than you can upload it, putting a server at home is something the broadband providers don't like very much because it consumes a lot of uplink broadband capacity. But with the passage of time, I believe it will be not only desirable, but quite natural, to have servers at home in addition to having use of the cloud. So symmetric capacity and IPv6 are a very attractive outcome and I'm not a big fan of carrier-grade NATs -- but it may turn out that NATs are needed in order to facilitate the transition during this period when we have to run both protocols. (See story: "Can Large Scale NAT Save IPv4?")
Do you think IPv6 in the home is as urgent as it is in the business network?
I think it's urgent in that if we don't get both protocols running at the same time, the day may come when there are servers that can only run IPv6, or there may be users who can only run IPv6 and couldn't get anything else, because the NAT boxes ran out.
I think it's important to get both protocols running smoothly at home. Already laptops and desktops have the capability. It's usually the firewall, the NAT box and maybe the broadband modem that you have at home that haven't been configured for IPv6. (See story: "Cisco Linksys routers still don't support IPv6")
So when we turn on IPv6 on a worldwide basis on June 8 as a 24-hour test (World IPv6 Day), I'm sure there will be things that don't work and those need to be addressed (no pun intended).
I would much rather see a concerted effort to get everybody up and running on IPv6. Then the transition is smooth because it doesn't matter if the destination is running IPv4 or 6, everyone can talk to everyone.
What do you see as the next major consumer use for the Internet?
If I knew that, I would be investing in it. But I think it's pretty clear that social networking is a big part of that. The platforms that are available that allow people to create their own content and share it is probably the most important avenue. The next one will be this Internet of Thingswhere we start managing collections of devices for our benefit. At some point third parties will figure out that they can come in and manage stuff for us. Our entertainment systems might be managed by a third party who doesn't have anything to do with the content, and simply provides organizing capacity, so we can make sure all of our entertainment material shows up on all the platforms where we want it.
You've publicly discussed your own home sensor network that alerts you when your wine cellar gets too hot or dry. You said it was built using technology from Arch Rock, acquired by Cisco in September. Is it Zigbee or Z-wave or something else?
It's actually 802.15.4, 6LoWPAN. It's a 6LoWPAN network running on IPv6.
The recent events in Egypt and the proposed U.S. government "Internet kill switch"... I get the sense that giving government the ability to turn the Internet off this is not something you would approve of. Is there a technological solution to prevent this?
Remember, the Internet runs on a substrate, and if the substrate is controllable by the government, it's possible to turn off the Internet. If we ever get to the point where mesh networking and let's say, peer-wise, point-to-point interactions can be done without benefit of things like routers provided by Internet service providers ... if you can build pieces of the Internet that sort of self-assemble [then the Internet could not be turned off].
Governments will control what they can control if they want to, and so the only solution around that is to have a network that is self-organizing. And technology is available to do that.
Julie Bort is the decided earthbound community editor for Network World. She pens the Odds and Ends blog for Cisco Subnet, the Microsoft Update blog for Microsoft Subnet and the Source Seeker blog for Open Source Subnet. Follow Julie on Twitter @Julie188
Read more about lans & wans in Network World's LANs & WANs section.

No comments: