IPv6: Enabling Mobility
By Justin Ried, Thu Nov 30 00:00:00 GMT 2000
Today's Internet addressing system (IPv4) is now more than 20 years old, and ISPs around the world are beginning to feel the pinch. In order to enable the mobile Internet we envision - one that offers the needed capacity, built-in security and mobility - we'll need to infuse modern technology into an aging infrastructure. And that can be tricky business.
Having a computer, or mobile phone, or anything at all connected to the Internet (called a node) requires that the device have an IP (Internet protocol) address. It's how other nodes differentiate between you and all other things on the network. Usually, your Internet service provider assigns an IP address to you dynamically when you logon. The problem today is that, like phone numbers, the number of addresses available on the Internet is shrinking fast - a process that will be sped-up by its mobilization.
So, in 1996 the Internet Engineering Task Force (IETF) began work on a new rewrite of the protocol, IPv6. Instead of the 32-bit addressing used in IPv4, the new version is based on a new, streamlined 128-bit structure. That means that instead of having only the 4 billion IP addresses presently available, we'll have a far more comfortable range of 30 Octillion (27 0s after the 30). With IPv4, many expect that we'll simply run out of addresses within 5-8 years. With the new system however, it doesn't seem likely we'll be running out of space anytime soon.
In addition to increasing the number of supported nodes, other benefits come with the new protocol. Streaming media, the kind to be used in third-generation phones, for example, stands to benefit greatly. Instead of simply dropping packets (or even cutting the connection) when the network becomes clogged, the quality of the picture will degrade. And when conditions improve, the quality of the picture does correspondingly.
How's it done? Well, if you've ever downloaded a JPEG image that has been "interlaced" you'll get the idea. When your browser makes a request for the image the whole outline and a very small amount of detail inside is sent back to you. As you continue to receive the data, it's pieced together in picture line-by-line, making it clearer and clearer as more and more bits arrive. Videos streamed to and from third-generation phones will operate in a similar fashion. The higher the throughput, the clearer the picture.
Security is another area where IPv6 benefits the end-user. Encryption and authentication markers are built right into the packet structure. Security will change from being an add-on to being an integral part of the network. This means that when a packet of sensitive information is sent from one node to another, only those two terminals will be able to decipher its contents. Contrast this to today's structure where the information must be encrypted/decrypted by the applications sending and receiving the data, and the data sent can be duplicated and decrypted without the owner being aware.
It works like this: When a packet is sensitive data is sent through an IPv6 network, the Encapsulating Security Payload (ESP) function encrypts everything except the header, which routes the packet to its destination. Once it's received, the other node decrypts the message using a key that was sent with the initial packet. "But what happens when the first packet gets it's key sniffed (or copied)," you might ask. Well, that's been considered also. Built into the IPv6 header is code that logs if and how many times the packet itself has been read. If the key has already been read by the time it reaches the legitimate recipient, it is discarded and a new one is sent.
Enhanced mobility is another key benefit. The current protocol was built around computers that rarely change location, if ever. When a mobile terminal is connected to the Internet today, it's forced to play by the same rules. The problem arises when a mobile phone is assigned an IP address by the operator, and then travels outside of that particular operator's coverage area and into another's. Remember, in third-generation networks, the phones are always connected via packet radio service, so there's going to be an issue with the handover.
IPv6 solves this by having the mobile terminal automatically inform the home network of its location and what the current shortest path to it is. The home network then routes the data accordingly. This way, the terminal can roam freely without changing its IP address. In a nutshell, IPv6 will identify each node by its static home address, regardless of it's current access point.
But given the clear benefits of IPv6's widespread adoption, some factors still stand in the way.
First and foremost is cost. Much of today's infrastructure will need to be replaced by IPv6 compatible equipment: Servers, routers, clients, and everything in between. Right now, IPv6 packets can be tunneled through existing equipment using what are called 6to4 routers to make the translation. These use an encapsulation technique similar to how the IPv6 encryption mechanism works.
But this is a temporary solution, and ideally, IPv6 packets won't need to go through such a transformation. In order to implement the technology on a wide scale, it'll need industry-wide support.
The number of companies backing the new protocol is impressive, and many seem to be taking action.
IBM has been offering compatibility since 1997, well ahead of anyone else. It was first offered in the OS/390 server platform. Soon, it'll be ready in OS/400, WebSphere, and Tivoli. Linux has had IPv6 in the kernel since version 2.2 came out, and Compaq's Tru64 UNIX has been compatible since September. Companies that have promised compatibility in the near-tern include Cisco, who has promised to ship all of their routers IPv6-enabled in the first quarter of next year. Sun Microsystems, who's servers power much of the 'net, promises to have IPv6 in every product next year also, including Solaris 8. Hewlett-Packard promises to build compatibility into its UNIX operating systems in 2001, and Apple says that Mac OS X (which is FreeBSD UNIX at its core) will have IPv6 code when it launches early next year.
Perhaps not surprisingly, Microsoft is a bit slow on the uptake. They have made a preview available to developers who may patch compatibility into Windows 2000 for testing purposes, but they say they won't build it into their commercial releases for at least two years.
The Third-Generation Partnership Project (3GPP) included IPv6 with the Release 2000 specification, so look for it to be standard on 3G handsets. In fact, all major handset manufacturers have pledged support, and Nokia just announced the first IPv6-compatible GPRS network this month.
How quickly the technology can be implemented remains to be seen, though. No one wants to be the first out there with a new, and relatively unproven technology. IT managers who write multi-million dollar checks are skeptical, since it's their job to maximize the return on their current infrastructure before buying into anything new. If they get out there too fast, they may encounter serious problems - which means free lessons on how to do it the right way for everyone else who waited.
Justin Ried still hasn't matched all the colors on his week-old Rubik's cube. But when he isn't trying to, he writes about mobile technology for TheFeature.